A system and method for integrated sensing and communication based on hybrid precoding fabric architecture

Abstract

This invention relates to a sensor-integrated system and method based on a hybrid precoded fabric architecture, comprising a metasurface element array, wherein the static phase states of all metasurface elements constitute a static precoded phase matrix; a column-based control network, including N independent control lines and a DC bias network, wherein each control line connects to and synchronously controls all metasurface elements in one column of the metasurface element array; a controller MCU, connected to the column-based control network, sending dynamic column control signals containing dynamic column control vectors to the column-based control network to generate the phase configuration of the final real-time beam scanning phase matrix and perform beamforming; and an external feed source for transmitting or receiving electromagnetic waves. The sensor-integrated system is configured to point the beam at a communication target in communication mode and perform beam scanning and process echo signals in sensing mode. This invention solves the phase error problem existing in RIS beam scanning systems and reduces the implementation complexity and manufacturing cost of smart fabrics.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication and intelligent sensing technology, and more particularly to a technology that combines reconfigurable smart surfaces (RIS) with smart fabrics. Specifically, it relates to a smart surface based on a hybrid precoded fabric architecture that integrates communication and sensing functions, and its control method. This technology aims to provide a low-cost, high-performance, and seamlessly integrated environmental intelligence platform for scenarios such as smart homes, environmental monitoring, and proactive security. Background Technology

[0002] Reconfigurable smart surfaces (RIS), also known as metamaterial surfaces, have attracted widespread attention in the field of communications due to their ability to intelligently control electromagnetic wave propagation in a low-power, low-cost manner, enabling functions such as beamforming, anomalous reflection, and focusing. While traditional active phased arrays offer superior performance, their expensive phase shifters and high power consumption limit their large-scale commercial deployment. Currently, most RIS implementations rely on traditional rigid printed circuit board (PCB) substrates. This rigidity makes it difficult to conformally integrate with everyday environments (such as furniture, walls, and floors) or the human body, limiting their application potential in emerging fields such as environmental intelligence and wearable devices. In terms of physical implementation, 1-bit RIS, requiring only two phase states (such as 0° and 180°), significantly reduces the complexity and cost of metasurface unit structures. However, in practical applications, RIS is typically non-uniformly illuminated by a single feed source, which introduces severe phase aberrations into the RIS aperture due to the spherical wavefront. This inherent phase mismatch, combined with the discrete phase error generated by 1-bit quantization, leads to problems such as beam pointing deviation and increased sidelobe levels, severely restricting the system's communication performance in wide target angle scanning.

[0003] Existing control schemes typically require independent control of each cell in the metasurface unit array. For an M×N RIS array, this necessitates M×N DC control lines and a complex wiring network, significantly increasing the complexity, cost, and power consumption of the hardware system. Providing independent DC bias for each metasurface cell requires complex wiring networks or multi-layer via processes, which not only increases manufacturing costs and assembly difficulty but may also introduce parasitic effects that interfere with RF performance. For large-area flexible substrates, traditional independent wiring schemes are virtually impossible to implement.

[0004] At the algorithmic level, beam control of a 1-bit reconfigurable smart surface is a high-dimensional, discrete, and non-convex optimization problem. Traditional mathematical optimization methods have high computational complexity, while heuristic algorithms converge slowly in high-dimensional search spaces, making it difficult to meet the requirements of real-time dynamic control.

[0005] Currently, wireless communication and wireless sensing (such as radar) are typically two separate systems. Communication systems prioritize high throughput and reliable data transmission, while sensing systems focus on high-precision target detection, localization, and imaging. This separate architecture leads to hardware resource redundancy, spectrum resource competition, and system integration complexity. Although academia has begun to explore the potential of RIS in the sensing field, existing technologies mostly focus on optimizing single functions, lacking an integrated hardware platform that can efficiently integrate both communication and sensing functions and operate stably in complex environments. Existing technologies still face significant challenges in achieving high-performance, multifunctional, and environmentally conformal smart surfaces.

[0006] In summary, existing technologies face three major challenges in achieving high-performance, multifunctional, and environmentally conformal smart surfaces: a lack of efficient solutions for morphological and functional integration; hardware manufacturing difficulties in realizing large-scale arrays on flexible substrates; and a lack of low-complexity, high-performance control algorithms that meet real-time requirements. These deficiencies limit the application of this technology in scenarios with stringent requirements for form, function, and performance, such as smart homes and health monitoring. Therefore, the core technical problem that this invention aims to solve is how to design a flexible, reconfigurable smart surface system that can achieve high-performance integrated communication and sensing with low computational complexity without significantly increasing hardware costs and manufacturing difficulties. Summary of the Invention

[0007] To address the issues of phase error, high computational complexity, and complex wiring of independent metasurface unit control lines in existing 1-bit RIS beam scanning systems, this invention provides a 1-bit RIS beam scanning optimization system and method based on a statistically driven hybrid precoded fabric architecture. The core idea of ​​this invention is to decompose the beam optimization process into two parts: a static precoded phase in the hardware and an online dynamic phase control via MCU signals. Statistical prior knowledge guides the pre-phase optimization, generating a static precoded phase matrix that compensates for the inherent phase difference between the conformal system's spherical wavefront and dynamic beam scanning. A low-complexity algorithm then enables rapid beam scanning. This aims to solve the problems of performance degradation of existing RIS systems under non-uniform spherical wave illumination, the computational complexity of high-dimensional optimization in 1-bit RIS systems, and the complexity of hardware control wiring.

[0008] The technical solution of this invention is as follows:

[0009] A synesthetic system based on a hybrid precoded fabric architecture includes a reconfigurable smart surface and an external feed. The reconfigurable smart surface comprises a columnar control network, a controller MCU, and a metasurface unit array based on a flexible textile substrate. Each metasurface unit in the array has a static phase state fixed during manufacturing. The static phase states of all metasurface units constitute a static precoded phase matrix. This static precoded phase matrix is ​​generated by a differential evolution algorithm based on the statistical laws of feed and beam scanning phases. It is used to pre-flip specific target phase units (those units that need to be flipped in the algorithm's pre-phase matrix optimization) to guide the final static phase state to be fixed in the metasurface unit array hardware, forming the final static precoded phase matrix and completing the pre-compensation for the inherent fixed phase aberration of the synesthetic system. The columnar control network has N control lines. Each control line connects to and synchronously controls all metasurface units in a column of the metasurface unit array, used to dynamically flip the static phase state; the controller MCU, connected to the column control network, is configured to calculate a dynamic column control vector using a greedy algorithm for the target direction, and send a dynamic column control signal containing the dynamic column control vector to the column control network; the final static precoded phase matrix is ​​combined with the dynamic column control vector calculated using the greedy algorithm based on the superposition of the static precoded phase matrix to generate the phase configuration of the final real-time beam scanning phase matrix, and beamforming is performed; the external feed cooperates with the reconfigurable smart surface space to transmit or receive electromagnetic waves; the integrated sensing system is configured to point the beam to the communication target in communication mode, and to perform beam scanning and process echo signals in sensing mode, and one structure can work in both communication and sensing modes.

[0010] Furthermore, each metasurface unit in the metasurface unit array includes an upper metal layer-insulator-lower metal layer structure. The upper metal layer includes a square bandgap structure and a PIN diode. Each square bandgap structure comprises two square bandgap structures symmetrically placed (i.e., symmetrical structure) along the control line direction connected to the metasurface unit, with adjacent sides being longer sides. A slot structure is left between the two adjacent sides to provide two stable reflection phase states (0° and 180°) and to allow DC biasing through the center of the slot. The PIN diode is placed at the center of the slot structure and spans the two square bandgap structures. The polarity direction of the PIN diode in the zero-phase state should ensure the polarity direction of the diodes in all metasurface units. The process is unified; the fixed static phase state of the metasurface unit is achieved by flipping the polarity direction of the PIN diode during manufacturing or by symmetrically flipping the geometric orientation of the metasurface unit with a square slot structure by 180 degrees during manufacturing; each metasurface unit includes an upper metal conductive cloth - a fabric dielectric substrate - a lower metal conductive cloth, and the square slot structure is a resonant patch; long rectangular microstrip lines are led out from the short sides of the two square slot structures for direct connection to the microcontroller DC-DC converter control network after the metasurface units are arranged; each metasurface unit includes a phase-flipping symmetrical structure design, and the DC bias network uses the phase-flipping symmetrical structure to simplify the DC bias line and is electrically connected to the PIN diode. This clearly points out that simplifying the DC bias line through the initial phase design is a key hardware structure, which improves the control accuracy of N lines controlling M×N metasurface units.

[0011] Furthermore, the columnar control network adopts a DC bias path design. Specifically, the DC bias network uses the symmetrical structure of the metasurface unit and the zero electric field generated by the double-sided slot structure at the operating frequency to introduce a DC bias signal. The slot structure extends the RF current path to generate high RF impedance, thereby achieving physical isolation between the DC control path and the RF resonant path, eliminating the need for multi-layer vias and independent wiring.

[0012] Furthermore, the static precoding phase matrix The phase flipping probability map is determined through an offline optimization process, which includes: performing statistical analysis within a preset target beam scanning angle range (e.g., 0 to 45 degrees with a step size of 5 degrees) to generate a flipping probability map characterizing the statistical law of phase flipping of each metasurface unit; using the initial phase matrix calculated based on the flipping probability map as the initial population, and employing a global heuristic optimization algorithm, preferably a differential evolution algorithm, to find a static phase matrix that optimizes the overall performance of the system within the target beam scanning angle range. The optimal overall performance is measured by maximizing an evaluation function, which is defined as the weighted sum of array gain, sidelobe level, and pointing accuracy at each discrete target angle within the scanning target angle range.

[0013] Furthermore, the evaluation function Quantification:

[0014]

[0015] in Indicates from the perspective of the target The peak gain of the array reflects the system's signal enhancement capability in that direction; This represents the highest sidelobe level. The lower the sidelobe level, the less signal interference there is and the stronger the system's anti-interference performance. Indicates the angle between the main lobe peak and the target. The deviation is used to measure the accuracy of beam pointing; The weighted phase error term between the static precoding phase matrix and the ideal phase distribution is calculated using the following formula:

[0016]

[0017] In the formula For the static precoded phase of the i-th metasurface unit, The reference ideal phase is obtained based on the statistical analysis of all scanned target angles. α is the weighting coefficient related to the flip probability of the i-th unit; α, β, and γ are preset fixed weighting coefficients, corresponding to the weighting proportions of the three indicators of gain, sidelobe, and pointing accuracy, respectively; η is the adaptive penalty weight related to the flip probability map, which can dynamically adjust the optimization strategy according to the phase flip statistical characteristics of each unit.

[0018] Furthermore, in the controller MCU, beamforming employs hybrid precoded beam control technology, including: acquiring a static precoded phase matrix fixed in the metasurface unit array; targeting the target beam pointing angle... A greedy algorithm is used to calculate the dynamic column control vector. For each column, the residual between the column and the ideal phase distribution is calculated when the entire column is flipped or not flipped based on the static precoded phase matrix seed. The state with the smaller residual is selected as the control value of the column. The optimized final static precoded phase matrix is ​​combined with the dynamic column control vector calculated by the greedy algorithm based on the static precoded phase matrix to generate the final real-time beam scanning phase matrix, and the metasurface unit array is driven by the column control network.

[0019] Furthermore, the reconfigurable smart surface array is integrated onto a flexible textile substrate; the flexible textile substrate is made of textile material with a relative permittivity of [missing information]. The loss tangent tanδ is 1.2 to 1.5, and the loss angle tangent tanδ is less than 0.005.

[0020] A method for implementing a synesthetic system based on a hybrid precoded fabric architecture, for realizing the synesthetic system based on a hybrid precoded fabric architecture as described above, includes the following steps:

[0021] S1. Based on the statistical phase requirements of the feed source within the preset scanning target angle range, a static precoding phase matrix is ​​generated through optimization using a differential evolution algorithm based on a hybrid precoding architecture. The static precoding phase matrix The fixed phase aberration required for pre-compensating the spherical wavefront generated by the feed source and the conformal reflecting surface of the flexible substrate deformation between the target angle required for beam scanning;

[0022] S2. The static precoding phase matrix Stored as the default phase state of the metasurface unit;

[0023] S3. Beam pointing angle for a target A dynamic column control vector C is calculated using a greedy algorithm. Each element of the dynamic column control vector C corresponds to a column of the metasurface unit array, and its value is 0 or 1, indicating whether an additional π phase flip is applied to the column.

[0024] S4. The static precoding phase matrix The dynamic column control vector C, calculated using a greedy algorithm and superimposed on the static precoded phase matrix, is combined to generate the final real-time beam scanning phase matrix. And thereby control each metasurface unit on the reconfigurable smart surface.

[0025] Furthermore, S3 specifically includes: for each column n in the metasurface unit array, calculating the residual Error_n between the column and the ideal phase distribution when π phase flipping is applied or not applied (i.e., k=0 or k=1), and selecting the state k that minimizes the residual Error_n as the dynamic control value of the column. The calculation formula is as follows:

[0026]

[0027] in It is the nth column of the static precoding phase matrix. Target angle The nth column of the corresponding ideal phase distribution.

[0028] A method for operating a synesthetic system based on a hybrid precoded fabric architecture, for realizing the synesthetic system based on the hybrid precoded fabric architecture as described above, includes:

[0029] In communication mode, for a specific communication user direction, a fixed dynamic column control vector is generated using the hybrid precoding beam control technology. This vector drives the metasurface element array to form a stable beam pointing towards the communication target. The system is then configured in communication mode, and the hybrid precoding beam control technology precisely points the beam towards the communication target, establishing or maintaining a communication link. In sensing mode, for a preset sensing target angle sequence or scanning area, the hybrid precoding beam control technology drives the metasurface element array to sequentially form a series of scanning beams pointing in different directions, scanning each target angle in the sequence. The external feed source receives echo signals reflected from the environment, and time-division multiplexing achieves integrated collaborative operation of communication and sensing functions. The beneficial effects of this invention are:

[0030] The reconfigurable smart surface, operating at 3 GHz, achieves precise beam control over a wide target angle scanning range of 0°-45° with a pointing error of less than 2°. Through columnar control, the number of required control lines is reduced from M×N to N. Combined with an innovative DC bias design, this significantly reduces the manufacturing difficulty and cost of realizing large-scale active arrays on flexible fabrics.

[0031] This application constructs a novel environmental intelligence platform by deeply integrating a hybrid precoding architecture, integrated sensing functionality, and fabric morphology, resulting in improvements in performance, practicality, and cost-effectiveness. This invention overcomes the core technical challenge of achieving high-performance integrated communication and sensing on flexible, large-area reconfigurable surfaces. Through a hybrid precoding fabric architecture, utilizing a static precoding phase matrix, this invention can pre-compensate for the phase difference between the feed near-field wavefront and the beam scanning target angle range, or for fixed phase aberrations introduced by the fabric's own physical deformation, ground coupling effects, and environmental static scattering, establishing a high-fidelity reference electromagnetic environment. Based on this, online dynamic column control calculated using a greedy algorithm based on the superposition of the static precoding phase matrix can generate precise beams in real time, thereby ensuring high gain and high stability of the communication link, as well as high resolution and high reliability of the sensing function, achieving a synergistic performance improvement for both.

[0032] This invention significantly reduces the complexity and manufacturing cost of smart fabrics, making large-scale commercial deployment feasible. Integrating a RIS array onto a large area of ​​fabric, using traditional independent metasurface unit control, would result in an extremely complex wiring network, almost impossible to implement. This invention's unique column-based control network reduces the number of control lines from M×N to N, fundamentally solving this hardware bottleneck. This simplified control architecture, combined with an innovative DC bias network design, makes the manufacturing of large-scale active arrays on flexible fabric substrates simple and economical. This invention opens a new paradigm for integrated sensing applications. Integrating sensing and communication functions into everyday household items like carpets has the potential to become an active sensing and interaction platform integrated into the environment. This invention is not merely a performance improvement of a single technology, but rather a collaborative innovation of software and hardware that moves environmental intelligence from concept to practical application. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the hybrid precoded fabric RIS array of the present invention.

[0034] Figure 2 This is a schematic diagram of the reconfigurable smart surface metasurface unit in an embodiment of the present invention.

[0035] Figure 3 The phase parameters S21 are the resonant simulation parameters of the metasurface unit in this embodiment of the invention.

[0036] Figure 4 This is the probability matrix seed for the static precoding phase matrix generated in the embodiments of the present invention.

[0037] Figure 5 This is a calculated radiation pattern under different scanning target angles in an embodiment of the present invention.

[0038] Figure 6 This is a schematic diagram of the final phase scheme in an embodiment of the present invention (taking 30 degrees as an example).

[0039] Figure 7 This is a flowchart of the algorithm in an embodiment of the present invention.

[0040] The attached diagram is labeled as follows: 1. External feed (speaker); 2. Metasurface array; 3. Column control network; 4. Controller; 5. Top resonant patch; 6. Fabric dielectric substrate; 7. PIN diode; 8. Bottom conductive cloth grounding layer. Detailed Implementation

[0041] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0042] A synesthetic system based on a hybrid precoded fabric architecture, such as Figure 1 As shown, it includes: an array of M rows and N columns of metasurface units integrated on a flexible textile substrate, wherein the physical structure of the metasurface units is designed to provide two different default static phase states, which are achieved by configuring the symmetry of the metasurface units and the physical orientation of the active elements integrated within them during the manufacturing stage; a column-based control network including N independent control lines, wherein each control line is electrically connected to all M metasurface units in a corresponding column of the array and is configured to synchronously control the phase states of all metasurface units in that column to dynamically flip; and a controller electrically connected to the column-based control network, configured to output control signals to the N control lines to dynamically adjust the phase states of the metasurface units based on their default static phases.

[0043] like Figure 2 As shown (the figure includes two metasurface cells), each metasurface cell in the array adopts a metal-insulator-metal (MIM) topology, consisting of, from top to bottom: two recessed rectangular top resonant patches, a dielectric substrate, and a complete bottom copper ground layer. Rectangular groove structures are etched on the left and right rectangular resonant patches of each metasurface cell. A PIN diode is connected across the center of the two recessed rectangular patches to achieve phase switching. Long rectangular microstrip lines are led out from both sides of the recessed rectangular resonant patches for direct connection to the microcontroller's DC-DC converter network after the cells are arranged. By arranging the two rectangular groove structures in a specific orientation (e.g., rotated 0° or 180°) during manufacturing or assembly, and accordingly determining the polarity of the PIN diodes, the metasurface cell is physically fixed to provide a default static phase response of 0° or 180° when no dynamic control signal is applied. The fixed static phase state of the metasurface cell is achieved by setting the polarity of the PIN diodes or the geometric orientation of the two rectangular groove structures during manufacturing.

[0044] The DC bias path of the column-type control network is preferably designed to introduce a DC signal by utilizing the electric field zero point generated by the double-sided slot structure at the operating frequency. The slot structure extends the RF current path to generate high RF impedance, thereby achieving physical isolation between the DC control path and the RF resonant path, eliminating the need for complex multi-layer vias. The column-type control network includes N independent control lines and a DC bias network. Each control line is electrically connected to all M metasurface units in a corresponding column of the array and is configured to synchronously control the phase state of all metasurface units in that column for dynamic flipping. The DC bias network utilizes the symmetrical structure of the metasurface units and the electric field zero point generated by the double-sided slot structure at the operating frequency to introduce a DC bias signal. The slot structure extends the RF current path to generate high RF impedance, thereby achieving physical isolation between the DC control path and the RF resonant path, eliminating the need for multi-layer vias and independent wiring.

[0045] The DC bias network introduces a DC bias signal by utilizing the electric field zero point, and increases the inductance by using the slot structure on both sides of the diode to extend the geometric path of the current distribution on the surface of the metasurface unit's RF signal. This results in high impedance at the operating frequency, thereby physically isolating the DC control path from the RF resonant path and eliminating the need for multi-layer vias and independent wiring.

[0046] The dielectric substrate is a flexible textile material with a relative permittivity of [missing information]. With a loss angle of 1.2 to 1.5, a loss tangent tan δ of less than 0.005, and a thickness of 6 mm, the selection of this textile substrate enables the metasurface unit array to possess conformal characteristics, allowing it to be seamlessly integrated into carriers such as carpets and walls, which is the key physical basis for realizing the integrated sensing function.

[0047] When in use, the reconfigurable smart surface requires an external feed horn for illumination. The feed horn is located at a distance of approximately 1000 mm from the center of the array along the normal direction. The spherical wavefront generated by the feed horn introduces a fixed phase aberration into the array aperture.

[0048] The controller is configured to execute a hybrid precoding method that supports both communication and sensing functions. In communication mode, the controller is configured to calculate the corresponding dynamic column control vector using a greedy algorithm based on the target azimuth angle of the communication user, and combine this with a static precoding phase matrix to precisely point the main lobe of the beam towards the communication user, thereby establishing a high-gain communication link for data transmission. In sensing mode, the controller uses a greedy algorithm to rapidly generate a series of dynamic column control vectors C according to a preset scanning strategy. Each vector C corresponds to a specific sensing scanning direction. By cyclically outputting these vectors in time sequence, the fabric surface formed by the arrangement of metasurface units can generate a series of scanning beams, performing point-by-point beam scanning of the covered space. During beam scanning, by receiving and analyzing the echo signals reflected from the human body or objects, the position, velocity, number, and even attitude of the target can be calculated, enabling advanced sensing functions such as fall detection, behavior recognition, and presence monitoring.

[0049] Its working principle is based on a unique hybrid precoding architecture, which cleverly decomposes the beam optimization process into two collaborative stages: offline static precoding and online dynamic control.

[0050] Step 1 (Offline Firmware and Hardware Implementation):

[0051] Each metasurface element in the metasurface element array is physically configured to a fixed static phase state (0 or π) during fabrication. The static phase states of all elements constitute a static pre-coded phase matrix. The matrix It is a compromise optimal solution obtained through an offline global optimization process and is permanently embedded in the hardware (metasurface unit). The quantitative definition of compromise optimality is: the static precoded phase matrix sought during the offline optimization stage. When this matrix is ​​combined with all subsequent possible dynamic column control vectors, it enables the system to achieve optimal overall performance across the entire preset scanning target angle range. This overall performance is evaluated using an evaluation function. Quantification:

[0052]

[0053] in Indicates from the perspective of the target The peak gain of the array reflects the system's signal enhancement capability in that direction. This represents the highest sidelobe level. The lower the sidelobe level, the less signal interference there is and the stronger the system's anti-interference performance. Indicates the angle between the main lobe peak and the target. The deviation is used to measure the accuracy of beam pointing. The weighted phase error term between the static precoding phase matrix and the ideal phase distribution is calculated using the following formula:

[0054]

[0055] In the formula For the static precoded phase of the i-th metasurface unit, The reference ideal phase is obtained based on the statistical analysis of all scanned target angles. η represents the weighting coefficients related to the flipping probability of the i-th unit. α, β, and γ are preset fixed weighting coefficients, corresponding to the weighting proportions of the three indicators: gain, sidelobe, and pointing accuracy, respectively. η is an adaptive penalty weight related to the flipping probability map, which can dynamically adjust the optimization strategy based on the phase flipping statistical characteristics of each unit. In the offline static precoding stage, the system performs in-depth analysis on the preset complete scan target angle range. By statistically analyzing the ideal phase requirements under each target angle, the system can generate a flipping probability map, which intuitively quantifies the state change tendency of each metasurface unit under different scanning scenarios. This probability map is not only important statistical prior knowledge but also a high-quality seed for subsequent optimization algorithms. Starting from this map, the system adopts heuristic optimizers such as differential evolution algorithms and introduces an adaptive penalty weight mechanism to guide the search process. In the fitness function design of the heuristic optimization algorithm, an adaptive penalty weight η is introduced. This weight is related to the unit statistical characteristics reflected by the flipping probability map and is used to dynamically adjust the optimization strategy. Its physical mechanism is based on the statistical analysis of the changes in unit phase requirements. When a cell has a high flip probability within a preset target scanning angle range, it indicates that the phase requirement of that cell changes drastically under different target scanning angles. These cells are usually located in the center region of the array, belonging to the phase-sensitive region that determines beam pointing accuracy. In this case, the algorithm automatically increases the value of η, forcing strict quantization error correction for these cells to ensure that the static pre-coded phase can provide an accurate reference phase and avoid pointing deviations caused by insufficient column control accuracy. Conversely, when a cell has a low flip probability, it indicates that its phase requirement is relatively stable. These cells are often located in the edge region of the array and mainly affect beam shape characteristics such as sidelobe levels. In this case, the algorithm will correspondingly decrease the value of η, allowing for beam shape optimization by fully utilizing the degrees of freedom of these cells, while moderately relaxing the phase accuracy constraints. For example, higher beam pointing gain can be obtained by sacrificing sidelobes, thereby improving the actual effective gain of the system.

[0056] The goal of optimization algorithms such as differential evolution is to maximize the above evaluation function. This adaptive adjustment mechanism based on statistical characteristics achieves an optimal physical balance between phase accuracy and beam performance optimization. The optimized static precoding phase matrix... Once directly embedded into the hardware, it ensures that the overall beam performance (including gain, sidelobe suppression, and pointing accuracy) of the system across the entire operating range is significantly better than that of traditional train control systems that do not employ such static precoding schemes.

[0057] It is worth emphasizing that the calculation result of this matrix does not merely remain at the algorithmic level; the final result of the algorithm optimization directly guides and is reflected in the hardware physical structure. By physically setting the polarity or geometric orientation of the switching elements (such as PIN diodes) in each metasurface unit during manufacturing, the optimized static phase matrix is ​​permanently embedded into the array's physical structure, replacing the original all-zero phase design. This process does not increase additional hardware costs but fundamentally pre-compensates for the inherent fixed phase aberrations of the system introduced by non-uniform illumination of the feed spherical wavefront and deformation of the flexible substrate, establishing a high-fidelity reference phase environment for subsequent real-time beam control. Through this adaptive adjustment based on statistical characteristics, the algorithm achieves a physically optimal balance between phase accuracy and beam performance optimization. The ultimate goal of this algorithm is not a simple intermediate solution but to calculate an "optimal" static precoded phase matrix. The "optimality" of this matrix is ​​reflected in the fact that when it is combined with subsequent dynamic control as a fixed base, it enables the system to achieve the best balance in overall beam performance across the entire scanning range.

[0058] Step Two (Online Dynamic Control and Beam Generation): The controller is configured to execute a hybrid precoded beam control method. For any target beam pointing angle... The controller uses a greedy algorithm to calculate a dynamic column control vector C in real time. This vector C is an N-dimensional binary vector, where each element represents whether the corresponding column needs to undergo a global π-phase flip. The system uses this vector to drive the array to form the real-time beam scanning phase matrix for the final beamformation. The phase matrix is ​​generated by XORing the static pre-encoded phase matrix and the dynamic vector C. This operation is performed in real time at each scanning target angle, achieving synergy between static precoding and dynamic control.

[0059] First, within the preset target angle range, a pure column control scheme using only dynamic column control without pre-phase is modeled. For each discrete target angle θ within a certain step interval within the scanning range, the theoretically required ideal continuous phase distribution is calculated using horn feed electromagnetic field theory. Next, Quantization is performed to a 1-bit ideal phase distribution (phase is 0 or π). The quantization process is to perform binarization mapping on continuous phase values.

[0060] During the online dynamic control phase, the system exhibits high flexibility and real-time performance. For any given target beam pointing at the target angle, the controller runs a column-level greedy algorithm with extremely low computational complexity. This algorithm makes independent decisions column by column, quickly calculating the required dynamic column control vector. Subsequently, this dynamic vector C is combined with the aforementioned static phase matrix already embedded in the hardware (usually through an XOR operation) to obtain the final real-time beam scanning phase matrix for the drive array to generate the target beam.

[0061] For each discrete target angle, a greedy algorithm based on pure column control is executed once to obtain the optimal column flip state vector at that target angle. In this algorithm, the decision variable k for each column represents the flip state of that column: k = 0 means that all cells in that column do not perform additional phase flips; k = 1 means that all cells in that column perform π-phase flips. Then, the frequency at which each metasurface unit is determined to need to be flipped (k = 1) within the entire preset scanning target angle range is statistically analyzed. After normalizing this frequency, the flipping probability value of the metasurface unit is obtained. The flipping probability values ​​of all metasurface units are used to form a flipping probability map. Then, a final static pre-coded phase matrix is ​​determined through subsequent heuristic global optimization algorithms. Before optimization, the final static pre-coded phase matrix corresponds to the static pre-coded phase matrix seed. The meaning of the static pre-coded phase matrix seed is that for units whose probability in the flipping probability map does not reach a certain statistical significance (e.g., threshold < 0.5), the final static pre-coded phase matrix is ​​suggested to set its initial phase to π (i.e., perform a 180-degree flip) in the hardware structure, and set the phase of units whose flipping probability reaches a sufficiently significant value to 0. In this way, the degree of freedom of control of units with high flipping requirements is given to the dynamic train control signal.

[0062] The aforementioned flip probability diagram physically reflects the statistical probability that each metasurface unit in the array, under all scanning target angles and in the optimal state derived from a pure column-control greedy algorithm, is required to flip to the π phase. The generation of this diagram simultaneously considers the phase compensation required for the far-field scanning target angles and the phase mismatch caused by the feed spherical wavefront illumination. This trade-off between the two phase requirements necessitates the π flipping of a subset of metasurface units initially in 0 phase.

[0063] Next, the static precoded phase matrix is ​​continuously optimized using a heuristic global optimization algorithm. During the optimization process, this matrix is ​​not yet fixed and serves only as a high-quality initial guess for seeking the global optimum. The phase distribution information corresponding to the above-mentioned static precoded phase matrix seed is used as a key input for subsequent heuristic global optimization algorithms (such as differential evolution) {logic chain: seed -> algorithm optimization -> output optimization result -> fixation}. The static precoded phase matrix seed is transformed into the initial population of the optimization algorithm, making it closer to the potential optimal solution, thereby accelerating the optimization process. A global heuristic optimization algorithm (such as differential evolution) is used to solve for the optimal final static precoded phase matrix. Unlike traditional completely random initialization, this invention uses the generated "flip probability map" as the "hot start" initialization population of the optimization algorithm, making the initial population closer to the global optimum, thereby accelerating convergence.

[0064] After the static precoding phase matrix seed is determined, a global heuristic optimization algorithm is used to calculate the final static precoding phase matrix. The first step is to calculate the dynamic column control state C required to generate the ideal phase state by superimposing the current static precoding phase matrix seed, starting with the target angle to be scanned. Calculate the ideal phase distribution at the target angle. Then, a low-complexity column-level greedy algorithm is used to calculate the dynamic column control matrix C. Specifically, for each column n (n=1, 2, ..., N) of the metasurface unit array, the phase residual is calculated under two dynamic states (no flip k = 0, flip k = 1):

[0065]

[0066] in, It is the nth column of the static precoding phase matrix. This is the nth column of the ideal phase matrix. Comparing Error_n under two states (k = 0 and k = 1), the state k that minimizes Error_n is selected as the dynamic control state for this column. By traversing all N columns, the complete N-dimensional dynamic column control vector can be obtained.

[0067] In the fitness function design of the heuristic optimization algorithm, an adaptive penalty weight η is introduced, which is related to the cell statistical characteristics reflected by the flip probability map. When the flip probability of a cell is high, it indicates that its phase requirement changes drastically with the target angle. In this case, α increases, and the global heuristic optimization algorithm focuses more on correcting the phase quantization error; conversely, α decreases, and the algorithm focuses more on maximizing the array gain. The optimization objective is to find a fixed final pre-encoded phase matrix such that, after combining with the subsequent greedy algorithm dynamic column control, for each target angle requiring beam scanning within the entire preset target angle range. Under this configuration, the RIS array gain and beam pointing accuracy (in terms of the main lobe peak direction and the target angle) corresponding to the real-time beam scanning phase matrix generated by superimposing the static precoded phase matrix with the dynamic train control phase matrix calculated using a greedy algorithm based on the static precoded phase matrix are shown. The weighted evaluation index (based on deviation measurement) is the highest. After this global optimization step, the final static precoded phase matrix is ​​the selected static precoded phase matrix that will be fixed as the hardware default state. Here, a performance constraint is introduced: the hardware configuration corresponding to any candidate final static precoded phase matrix, after combining dynamic column control phase, must have a system gain better than the pure column control phase scheme without precoding. Finally, a globally optimal static initial phase distribution is determined, which guides the physical structure design of the metasurface unit.

[0068] The final static precoded phase matrix that needs to be solidified in the hardware is combined with the column-level flip operation represented by the real-time dynamic column control vector calculated using a greedy algorithm based on the static precoded phase matrix to generate the final real-time beam scanning phase matrix. Then, based on the dynamic column control vector calculated using a greedy algorithm based on the static precoded phase matrix, the final static precoded fabric phase system solidified in the column control hardware performs dynamic column vector flip operation to drive each metasurface unit of the reconfigurable smart surface.

[0069] The final static precoded phase matrix generated by the algorithm is used in hardware manufacturing, and the microcontroller in the system is used to output column control signals. The final hardware RIS phase configuration matrix. By using the final static precoding phase matrix The final real-time beam scanning phase matrix is ​​obtained by superimposing the dynamic column control matrix C, calculated using a greedy algorithm based on the static precoded phase matrix, with the phase matrix superimposed on the dynamic column control matrix C (equivalent to an XOR operation). The controller generates N DC control signals based on the dynamic column control state C calculated by the greedy algorithm, and drives the phase configuration of each metasurface unit on the reconfigurable smart surface through the column DC bias control network, thereby completing the fast and precise deflection of the beam.

[0070] This invention relates to a sensor-integrated system based on a hybrid precoded fabric architecture, a product invention whose hardware structure comprises four parts: an external feed source, a metasurface element array, a columnar DC bias control network, and a controller. These components work collaboratively to achieve efficient beam-scanning electromagnetic wave modulation and are applied in real-world sensor-integrated scenarios.

[0071] This invention discloses a sensory integrated fabric system and its control method based on a hybrid precoding architecture, belonging to the field of wireless communication technology. This technology is mainly applied in wireless communication scenarios such as satellite communication terminals, beam scanning, and sensory integration. This invention aims to solve the problems of performance degradation of existing RIS under non-uniform spherical wave illumination, high computational complexity of high-dimensional discrete optimization, and complex hardware wiring techniques. This invention constructs a system including an external feed source, an optimized metasurface element array with partially flipped metasurface elements, and a simplified columnar control network. The core technical solution of this invention aims to build a high-efficiency, low-cost beamforming system. It consists of an optimized reconfigurable smart surface (RIS) array and a simplified columnar control network. Its working principle is based on a unique hybrid precoding architecture, which cleverly decomposes the beam optimization process into two collaborative stages: offline static precoding and online dynamic control.

[0072] Therefore, the hybrid precoding method of this invention achieves deep coupling between the algorithm and hardware. In the offline stage, a statistically driven method is used to generate and solidify the optimal static phase matrix; in the online stage, a simplified greedy algorithm is relied upon for rapid real-time decision-making. This collaborative mechanism brings several significant advantages: at the hardware level, by combining solidified static phase with column-based control, the complex architecture that originally required independent wiring for each unit (M×N lines, such as 196 lines) is simplified to a concise design that only requires controlling each column (N lines, such as 14 lines), reducing the number of control lines by approximately 93%, which is particularly crucial for achieving large-scale arrays on flexible fabric substrates. At the algorithm level, the real-time computational complexity is significantly reduced from O(M×N) to O(N). Finally, in terms of performance, the system achieves excellent performance with a beam pointing error of less than 2° and effective sidelobe level suppression within a wide target angle scanning range of 0° to 45°.

[0073] In summary, this invention, through the aforementioned collaborative hardware and software innovation, systematically overcomes several major challenges commonly encountered in wide-target-angle scanning of existing 1-bit reconfigurable smart surface systems. These challenges include beam pointing inaccuracies and increased sidelobes caused by phase quantization and spherical wavefront distortion; the high computational cost of traditional optimization methods, making them difficult to apply in real-time; and the extremely complex hardware wiring required for large-scale array control. Unlike conventional circuit boards, this invention uses fabric and utilizes the phase-flipping method of the initial array to reduce the number of DC power supply control lines from M×N to N without increasing hardware costs, thereby reducing real-time computational complexity. Simultaneously, it achieves excellent performance with a pointing error of less than 2° within the 0°-45° scanning range. This invention aims to address the technical shortcomings of existing 1-bit RIS systems in wide-target-angle scanning, such as phase quantization errors, beam pointing inaccuracies caused by non-uniform spherical wavefront illumination, high sidelobes, and the high computational complexity and complex hardware wiring of traditional optimization methods.

[0074] Example 1

[0075] Please see Figure 1As shown, this embodiment first provides a synesthetic integrated fabric system based on a hybrid precoding architecture. The system includes an external feed horn, a fabric metasurface element array 2 located below the feed horn, where each metasurface element has a static phase state fixed during manufacturing. The static phase states of all elements constitute a static pre-coded phase matrix. This matrix is ​​generated using an optimization algorithm and ultimately fixed in the metasurface hardware to pre-flip specific phase elements, pre-compensating for inherent fixed phase aberrations. A column control network 3 has N control lines, each connecting and synchronously controlling all metasurface elements in a column of the array for dynamic phase state flipping. A controller 4, connected to the column control network, is configured to calculate a dynamic column control vector for a target direction. The final static pre-coded phase matrix is ​​combined with the dynamic column control vector calculated using a greedy algorithm based on this matrix to generate the final real-time beam scanning phase matrix for beamforming. A feed horn works in conjunction with a reconfigurable smart surface space. The column control network connects the metasurface and the controller MCU, receiving dynamic column control signals from the controller. The system is configured to point the beam at a communication target in communication mode and perform beam scanning and process echo signals in sensing mode. In a preferred embodiment, the communication and sensing modes work together using time-division multiplexing. The controller (MCU) divides the time axis into repeating frame structures, each frame containing one communication time slot and one or more sensing scan time slots. Within the communication time slot, the system uses hybrid precoding beam control technology to point the beam at a known communication user for data transmission; within the sensing time slot, the system sequentially points the beam at different target angles according to a preset scanning sequence to perform environmental sensing and receive echo signals. The length ratio of the communication time slot to the sensing time slot can be dynamically configured according to requirements.

[0076] In this embodiment, the flexible textile substrate is preferably applied to carpet products. Using carpet as the application carrier fully utilizes its large-area coverage in everyday home environments, enabling wide-range electromagnetic wave modulation and sensing coverage. Furthermore, as a natural floor support in indoor environments, carpet facilitates the spatial arrangement of the feed source and RIS array, and is less prone to obstruction, resulting in better perception of human movement. It should be noted that, in addition to carpet, the flexible textile substrate can also be applied to other home textiles such as wall coverings, all of which are equivalent embodiments of the present invention. The system operates in the 2.95 GHz frequency band, with a metasurface unit period of 48 mm. There is a spatial positional relationship between the feed source 1 and the metasurface unit array 2. The feed source 1 is located at a specific distance (i.e., focal length Fz) on the center normal of the metasurface unit array 2 to form spherical wavefront illumination. The controller 4 is connected to the column control network 3 via electrical signals. The column control network 3 is electrically connected to each column of metasurface units in the fabric metasurface unit array 2 through a DC bias microstrip line structure on the fabric surface.

[0077] Please see Figure 2 As shown, Figure 2 This is a schematic diagram showing two fabric metasurface units connected via a DC microstrip line structure. The reconfigurable smart surface metasurface unit in this embodiment adopts a metal-insulator-metal (MIM) topology. Each metasurface unit, from top to bottom, includes: two rectangular top resonant patches 5 with grooves, a fabric dielectric substrate 6, and a complete bottom conductive ground layer 8. Rectangular groove structures are etched on the left and right resonant patches of each unit. A PIN diode 7 is connected across the center of the two rectangular groove structures. A forward or reverse bias voltage is applied through a columnar control network 3, allowing the metasurface unit to switch between the two states, thus achieving a reflection phase difference of approximately 180°. When the controller 4 applies a forward bias voltage to the PIN diode 7 through the columnar control network 3, the diode is in the ON state, equivalent to a low-impedance RL circuit, and the metasurface unit array exhibits a 0° phase reflection. When a reverse bias voltage is applied, the diode is in the OFF state, equivalent to a high-impedance RC circuit, and the metasurface unit array exhibits a 180° phase reflection. By controlling the state of PIN diode 7, 1-bit quantitative control of the phase of the reflected electromagnetic wave can be achieved.

[0078] In this embodiment, the fabric dielectric substrate 6 is made of a low-loss, low-cost flexible textile material with a thickness h of approximately 3-6 mm and a relative permittivity. The loss tangent tanδ is approximately 1.2, and the loss angle tangent tanδ is less than 0.005. The key geometric parameters of the top resonant patch 5 are: external dimensions: the outer perimeter of each rectangular patch has a length L1 of approximately 30 mm and a width W1 of approximately 20 mm; the inner groove has a length L2 of approximately 25 mm and a width W2 of approximately 4 mm.

[0079] One of the hardware innovations of this embodiment lies in its simplified DC bias network. While controlling the symmetrical metasurface cells in a single row, the diodes of some metasurface cells are placed in reverse to achieve a phase rotation effect, compensating for the static phase difference of the conformal reflective surface caused by spherical feed or flexible substrate deformation. For example... Figure 2 As shown, at the operating frequency, the center of the rectangular patch near the short side of the slot is the electric field zero point. This structure utilizes this electric field zero point and the slot design to extend the propagation path of electromagnetic waves on the surface of the rectangular patch, reducing the interference of RF signals on DC signals. Therefore, a rectangular microstrip line structure is introduced near the short side of the slot of the rectangular patch to transmit the DC bias signal, presenting a high impedance state to the RF signal at the target frequency. This design achieves physical isolation between the DC bias path and the RF resonant path, eliminating to some extent the interference of DC bias lines and vias on electromagnetic performance in traditional designs, and can be achieved with only a single-layer dielectric process.

[0080] Figure 3 The figure shows the S-parameters of the metasurface unit in the embodiment of the present invention for resonance simulation. As can be seen from the figure, when the PIN diode is in the 'OFF' (reverse bias) and 'ON' (forward bias) states, the metasurface unit achieves a reflection phase difference of approximately 180°. This result verifies the effectiveness of the metasurface unit as a 1-bit encoding device.

[0081] To achieve high-performance beam control under wide target angle scanning, this invention employs a hybrid precoding method. This method decouples the beamforming process into physical offline static precoding and online dynamic control. Controller 4 executes the hybrid precoding method to generate a 14×14 static precoding phase matrix. This matrix is ​​initialized using the initial phase matrix calculated based on the flip probability map within the 0°-45° scanning range, and optimized using a differential evolution algorithm with adaptive penalty weights. This matrix is ​​used to pre-compensate the static phase difference between the spherical wavefront of the feed horn 1 and the beam deflection scanning, and is stored as a hardware default state. Its heatmap relative to the required metasurface unit flip probability of the dynamic column is shown below. Figure 4 As shown.

[0082] Please see Figure 5 As shown, it illustrates the calculated radiation patterns of the present invention at different scanning target angles (0°, 15°, 30°, 45°). In contrast, the "control only" curve represents the curve without static precoding (i.e.,...). The performance at 30° and 45° shows a gain decrease. However, by adopting the hybrid precoding method of this invention, the beam scanning gain is improved compared to the pure column control array, demonstrating the effectiveness and superiority of this invention within the 0°-45° scanning range.

[0083] Please see Figure 6 The diagram illustrates a scheme for obtaining the final real-time beam scanning phase matrix by XORing the final static precoded phase matrix with a dynamic column control matrix calculated using a greedy algorithm based on the superposition of the static precoded phase matrix. This is useful when the beam needs to be pointed at a specific target angle, such as... At that time, for the final static precoding matrix that has been fixed in the hardware and optimized by the heuristic algorithm, controller 4 first executes a greedy algorithm to calculate a dynamic column control vector C, whose corresponding binary encoding matrix is ​​as follows: Figure 6 The dynamic column control matrix is ​​shown. Then, controller 4 performs an XOR operation. The final phase configuration is obtained, and the result is as follows: Figure 6 The final real-time beam scanning phase matrix is ​​shown. Finally, controller 4 outputs a control signal based on this vector C, and performs columnar control of the array through 14 control lines to achieve rapid beam deflection.

[0084] In summary, the RIS and its control method based on a hybrid precoding framework proposed in this invention decomposes the complex online optimization problem into offline precoding and online simplified control through collaborative innovation of software and hardware. Combined with an innovative columnar control network hardware architecture, it achieves low-cost, low-complexity, and high-performance real-time beam control, providing an efficient and feasible solution for the practical application of RIS in intelligent wireless environments.

[0085] Alternative technical solutions

[0086] Alternative to the metasurface unit array structure: In this invention, the metasurface unit employs a slot structure that can achieve 1-bit phase modulation via DC bias. As a preferred embodiment, this slot structure is a double-sided slot. It is understood that this core design concept is also applicable to other slot structures that can provide two stable reflection phase states (0° and 180°) and allow DC bias through the slot center, such as, but not limited to, "H"-shaped, "I"-shaped, or Jerusalem cross-shaped slots. When using slot structures of different shapes, their geometric parameters need to be adjusted accordingly to meet phase and impedance matching requirements.

[0087] Algorithm Alternatives: In the offline computation stage of the static precoding phase matrix, this invention employs a heuristic optimization algorithm for global optimization. In a preferred embodiment, the differential evolution algorithm is used as the optimizer. It is understood that other heuristic algorithms, such as binary particle swarm optimization (BPSO), binary gray wolf optimizer (BGWO), or simulated annealing (SA), can also be applied to this optimization problem. The "hot start" initialization strategy and adaptive penalty weight mechanism based on the flipped probability graph proposed in this invention, as a general framework improvement, are also applicable to the aforementioned alternative algorithms to improve their convergence speed and final solution quality.

[0088] Alternative Control Granularity: In the real-time beamforming stage, this invention employs a mechanism of grouping and controlling metasurface units to achieve a balance between performance and system complexity. In a preferred embodiment, the control granularity is set to "column-level," meaning each column of metasurface units shares a single dynamic control signal. It is understood that this control granularity can be adjusted according to the specific requirements of different application scenarios regarding the trade-off between accuracy and complexity. For example, the initial phase of the unit can be adjusted using the symmetry of the unit itself, further dividing adjacent units into blocks (such as 2×2 or 4×4 units as a control block) for "block-level" control. Block-level control further reduces the number of control lines required, but may introduce some beamforming accuracy loss. This granularity adjustment under the grouping control concept is considered a simple variant of the equivalent concept of this invention.

[0089] This invention integrates synesthetic functionality onto flexible fabric. It's not simply placing a rigid RIS (Resonance Information System) on the fabric; rather, it's an integrated design encompassing hardware structure, control algorithms, and application modes, creating a completely new environmental intelligence platform that seamlessly blends with the environment. This is the fundamental difference between this invention and all existing RIS technologies.

[0090] Key Protection Point 1 (Hardware Architecture): The columnar control network and its integration with the fabric substrate. Please ensure the protection of the hardware architecture where N control lines control an M×N array. This architecture is the physical basis for realizing large-scale, flexible arrays, solving the fundamental problem of traditional independent wiring being impossible on fabric.

[0091] Key Protection Point Two (Control Method): Hybrid Precoding Framework. This is the core algorithmic idea for achieving high-performance, low-complexity control. It is recommended to break it down into independent, corresponding static parts of the method steps: Protecting the fixed static precoding phase matrix ( The process of pre-compensating for fixed phase aberrations such as flexible substrate deformation and ground coupling is particularly important. The dynamic component protects a low-complexity greedy algorithm for calculating the dynamic column control vector online in real-time, based on the static pre-encoded phase matrix superimposed with a greedy algorithm, and how this vector is combined with the static matrix. Synergistic effect: The static and dynamic components are combined to generate the final real-time beam scanning phase matrix. The complete method.

[0092] Key Protection Point Three (Application Mode): Integrated Sensing and Communication Working Mode. The same hardware platform reduces the search space for conformal fabric beam phase calculations for flexible substrate deformation through simple switching control logic, facilitating the separate or simultaneous implementation of communication beamforming and sensing beam scanning functions. This demonstrates the system integration value and application prospects of this invention.

[0093] Appendix Figure 1 In this configuration, N control lines are connected one-to-one with the DC bias microstrip lines in each column. The control signal transmission path includes the signal line and the return ground line. Because the metasurface unit adopts a symmetrical structure and a common ground design on both sides, the return ground lines of each column can share a single common ground line connected to the controller ground terminal. Therefore, each column has one effective control line, and N columns require a total of N control lines. In this embodiment, N=14, for a total of 14 control lines.

[0094] The above-described embodiments are merely one implementation of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A system for integrated sensing and communication based on hybrid precoding fabric architecture, comprising: The system includes a reconfigurable smart surface and an external feed source. The reconfigurable smart surface comprises a columnar control network, a controller MCU, and a metasurface unit array based on a flexible textile substrate. Each metasurface unit in the array has a static phase state that is fixed during manufacturing. The static phase states of all metasurface units constitute a static pre-coded phase matrix. This static pre-coded phase matrix is ​​generated and fixed in the metasurface unit array based on the phase statistics of the feed source and beam scanning using a differential evolution algorithm. It is used to pre-compensate for the inherent fixed phase aberration of the integrated sensing and communication system. The columnar control network includes N independent control lines and a DC bias network. Each control line is connected to and synchronously controls the external feed source. All metasurface units in a column of the metasurface unit array are used to dynamically flip the static phase state; the controller MCU, connected to the column control network, is configured to execute a greedy algorithm to calculate a dynamic column control vector for the target direction, and send a dynamic column control signal containing the dynamic column control vector to the column control network; the final static precoded phase matrix is ​​combined with the dynamic column control vector to generate the phase configuration of the final real-time beam scanning phase matrix, and beamforming is performed; the external feed cooperates with the reconfigurable smart surface space to transmit or receive electromagnetic waves; the integrated sensing system is configured to point the beam to the communication target in communication mode, and to perform beam scanning and receive echo signals in sensing mode.

2. The synesthetic system based on a hybrid pre-coded fabric architecture according to claim 1, characterized in that, Each metasurface unit in the metasurface unit array includes an upper metal layer-insulator-lower metal layer structure. The upper metal layer includes a square bandgap structure and a PIN diode. Each square bandgap structure comprises two square bandgap structures symmetrically placed along the control line direction connected to the metasurface unit, with adjacent sides being longer. A slot structure is left between the two adjacent sides to provide two stable reflection phase states (0° and 180°) and allow DC biasing through the center of the slot. The PIN diode is placed at the center of the slot structure and spans the two square bandgap structures. The polarity direction of the PIN diode in the zero-phase state should ensure that the polarity direction of the diodes in all metasurface units is uniform. The fixed static phase state of the metasurface unit is achieved by flipping the polarity direction of the PIN diode during manufacturing or by symmetrically flipping the geometric orientation of the metasurface unit with the square slot structure by 180 degrees during manufacturing. Each metasurface unit includes an upper metal conductive cloth, a fabric dielectric substrate, and a lower metal conductive cloth, with the square slot structure serving as a resonant patch. Long rectangular microstrip lines are led out from the short sides of both square slot structures for direct connection to the microcontroller DC-DC converter network after the metasurface units are arranged. Each metasurface unit includes a phase-flipping symmetrical structure design, and the DC bias network uses the phase-flipping symmetrical structure to simplify the DC bias line and is electrically connected to the PIN diode.

3. The system of claim 1, wherein, The DC bias network utilizes the symmetrical structure of the metasurface unit and the zero electric field generated by the bicornuate slot structure at the operating frequency to introduce a DC bias signal. It also extends the RF current path through the slot structure to generate high RF impedance, thereby achieving physical isolation between the DC control path and the RF resonant path and eliminating the need for multi-layer vias and independent wiring.

4. The system of claim 1, wherein, The static precoded phase matrix The determination is achieved through an offline optimization process, which includes: performing statistical analysis within a preset target beam scanning angle range to generate a phase reversal probability map characterizing the statistical law of phase reversal for each metasurface unit; using the initial pre-encoded phase matrix calculated based on the phase reversal probability map as the initial population, and employing a global heuristic optimization algorithm, preferably a differential evolution algorithm, to find a static phase matrix that optimizes the overall performance of the system within the target beam scanning angle range. The optimal overall performance is measured by maximizing an evaluation function, which is defined as the weighted sum of array gain, sidelobe level, and pointing accuracy at each discrete target angle within the scanning target angle range. In the fitness function design of the heuristic optimization algorithm, an adaptive penalty weight η is introduced, which is related to the unit statistical characteristics reflected by the flip probability map and is used to dynamically adjust the optimization strategy.

5. The system according to claim 4, wherein, Evaluation function Quantification: in Indicates from the perspective of the target The peak gain of the array reflects the system's signal enhancement capability in that direction; This represents the highest sidelobe level. The lower the sidelobe level, the less signal interference there is and the stronger the system's anti-interference performance. Indicates the angle between the main lobe peak and the target. The deviation is used to measure the accuracy of beam pointing; The weighted phase error term between the static precoding phase matrix and the ideal phase distribution is calculated using the following formula: In the formula For the static precoded phase of the i-th metasurface unit, The reference ideal phase is obtained based on the statistical analysis of all scanned target angles. α is the weighting coefficient related to the flip probability of the i-th unit; α, β, and γ are preset fixed weighting coefficients, corresponding to the weighting proportions of the three indicators of gain, sidelobe, and pointing accuracy, respectively; η is the adaptive penalty weight related to the flip probability map, which can dynamically adjust the optimization strategy according to the phase flip statistical characteristics of each unit.

6. The system of claim 1, wherein, In the controller MCU, beamforming employs a hybrid precoded beam control technique, including: obtaining a static precoded phase matrix embedded in the metasurface unit array using a statistically driven differential evolution algorithm; and targeting the target beam pointing angle. A greedy algorithm is used to calculate the dynamic column control vector. For each column, the residual between the column and the ideal phase distribution is calculated when the entire column is flipped or not flipped based on the static precoded phase matrix seed. The state with the smaller residual is selected as the control value of the column. The optimized final static precoded phase matrix is ​​combined with the dynamic column control vector calculated by the greedy algorithm based on the static precoded phase matrix to generate the final real-time beam scanning phase matrix, and the metasurface unit array is driven by the column control network.

7. The system of claim 1, wherein, The reconfigurable smart surface array is integrated on a flexible textile substrate; the flexible textile substrate is made of textile material with a relative permittivity. The loss tangent tanδ is 1.2 to 1.5, and the loss angle tangent tanδ is less than 0.

005.

8. An implementation method of a mixed precoding fabric architecture based C4ISR integrated system, characterized in that, To implement the synesthetic system based on a hybrid precoded fabric architecture as described in any one of claims 1-7, the method includes the following steps: S1. Based on the statistical phase requirements of the feed source within the preset scanning target angle range, a static precoding phase matrix is ​​generated through optimization using a differential evolution algorithm based on a hybrid precoding architecture. The static precoding phase matrix The conformal reflecting surface used to pre-compensate for the spherical wavefront or flexible substrate deformation generated by the feed source requires a fixed phase aberration between the target angles required for beam scanning. S2. store the static precoding phase matrix as the default phase state of the metasurface unit; S3. Beam pointing angle for a target A dynamic column control vector C is calculated using a greedy algorithm. Each element of the dynamic column control vector C corresponds to a column of the metasurface unit array, and its value is 0 or 1, indicating whether an additional π phase flip is applied to the column. S4. The static precoding phase matrix The dynamic column control vector C, calculated using a greedy algorithm and superimposed on the static precoded phase matrix, is combined to generate the final real-time beam scanning phase matrix. And thereby control each metasurface unit on the reconfigurable smart surface.

9. The implementation method of the mixed precoding fabric based C4I system according to claim 8, characterized in that, S3 specifically includes: for each column n in the metasurface unit array, calculating the residual Error_n between the column and the ideal phase distribution when π phase flipping is applied or not applied (i.e., k=0 or k=1), and selecting the state k that minimizes the residual Error_n as the dynamic control value Cn for the column. The calculation formula is as follows: in, It is the nth column of the static precoding phase matrix. Target angle The nth column of the corresponding ideal phase distribution.

10. A working method of a mixed precoding fabric based system of common sense integration, characterized in that, For implementing the synesthetic system based on a hybrid precoded fabric architecture as described in claim 6, the system includes: In communication mode, for a specific communication user direction, a fixed dynamic column control vector is generated using the hybrid precoding beam control technology. This vector drives the metasurface element array to form a stable beam pointing towards the communication target. The system is then configured in communication mode, and the hybrid precoding beam control technology precisely points the beam towards the communication target, establishing or maintaining a communication link. In sensing mode, for a preset sensing target angle sequence or scanning area, the hybrid precoding beam control technology drives the metasurface element array to sequentially form a series of scanning beams pointing in different directions, scanning each target angle in the sequence. The external feed source receives echo signals reflected from the environment, and time-division multiplexing enables integrated collaborative operation of communication and sensing functions.

Images