Parking space state detection method based on RIS auxiliary imaging

Abstract

The invention discloses a parking space state detection method based on reconfigurable intelligent super surface (RIS) auxiliary imaging, and belongs to the technical field of wireless sensing. According to the method, imaging of a target area (ROI), namely a parking lot, is completed through cooperation of multiple RISs, and real-time monitoring of the parking space state is achieved. Firstly, before detection is started, if no vehicle exists in a parking lot, a transmitting antenna transmits sensing signals for multiple times, a receiving antenna extracts channel components of the signals sequentially passing through an ROI and an RIS, and the channel components serve as reference channel measurement corresponding to the concrete ground. And then detecting the parking lot, repeating the process to obtain the measurement of the parking lot at the moment, subtracting the reference measurement to obtain a channel difference value, and capturing the influence caused by the possible existence of the vehicle. Then, an imaging target is divided into a plurality of grid units, the scattering coefficient of each unit is an imaging solving target, and a sensing matrix is generated based on known parameters; and finally, imaging is realized by using a compressed sensing algorithm, and the occupation state of the parking space is judged based on a result.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless sensing, and particularly relates to a method for detecting the state of a parking space based on RIS-assisted imaging. Background Technique

[0002] With the acceleration of the urbanization process and the continuous growth of the automobile ownership, the problem of parking difficulty has become increasingly prominent. Real-time and accurate detection of the state of parking spaces in a parking lot is of great significance for improving parking efficiency and alleviating traffic congestion. Traditional parking space detection schemes mainly rely on devices such as intelligent cameras or geomagnetic sensors, but these schemes have obvious deficiencies. Intelligent cameras require high-resolution image processing and have high requirements for lighting and environmental conditions; geomagnetic sensors need to lay sensors above the parking space, increasing the deployment cost and complexity. In contrast, detecting the state of a parking space based on RIS-assisted imaging with wireless sensing as the core has advantages such as flexible deployment, low cost, and strong environmental adaptability, providing a new solution for parking space state detection.

[0003] In wireless sensing technology, imaging technology can provide richer environmental information than simple positioning and is an effective means to realize the detection of the state of a parking space. As a new type of electromagnetic wave modulation device, the Reconfigurable Intelligent Surface (RIS) plays an important role in wireless imaging. The RIS consists of a large number of programmable reflection units, and by modulating parameters such as the amplitude and phase of electromagnetic waves, precise control of the spatial beam can be achieved. Compared with traditional wireless relays, the RIS has advantages such as small volume, low cost, and high energy efficiency, and can also avoid introducing self-interference and support full-duplex transmission. These characteristics make the RIS have broad application prospects in the field of wireless imaging.

[0004] However, there are still some limitations in the current RIS-assisted imaging technology. Existing research is mostly limited to the application of a single RIS and fails to fully utilize the potential of multi-RIS collaborative work. The coverage range of a single RIS is limited, and the imaging accuracy needs to be improved. Especially in a complex parking lot environment, it is difficult to meet the requirements of high-precision parking space state detection. In addition, the existing algorithm design and system verification are not yet mature, which limits the actual application effect of the RIS in parking space detection.

[0005] Based on the above analysis, the present invention proposes a method for detecting the state of a parking space based on RIS-assisted imaging. This method deploys multiple RIS units to construct a collaborative imaging system to expand the detection coverage range and improve the imaging accuracy. By optimizing the RIS reflection parameter configuration and imaging algorithm, precise perception of the parking lot environment and reliable detection of the state of the parking space are realized. Compared with traditional schemes, the present invention has advantages such as flexible deployment, low cost, and strong environmental adaptability, providing a new technical solution for intelligent parking lot management. Summary of the Invention

[0006] Object of the Invention: Aiming at the problems of high cost and complex deployment in the current parking space status detection, this solution proposes a RIS-assisted parking space status detection method. By utilizing the electromagnetic wave regulation ability of RIS, several RISs can be arranged to cover a small parking lot. Compared with traditional intelligent cameras and geomagnetic sensors, this solution has significant cost advantages and simplifies device deployment. With the efficient reflection and adjustment functions of RIS, the occupancy status of parking spaces can be detected in real time. This system not only reduces the operating cost of parking lot management but also improves the parking experience of car owners, realizing efficient and accurate parking space management.

[0007] Technical Solution: A RIS-assisted parking space status detection method of the present invention uses several reconfigurable intelligent surfaces (RISs) to cooperate to complete communication and imaging of the region of interest (ROI), and judge the status of the ROI unit (whether there is a car). The steps for realizing its functions are as follows:

[0008] Step 1: Install multiple RISs at different positions several meters above the ground in the parking lot to construct a distributed RIS system, and install a transmitting antenna and a receiving antenna at appropriate positions in the parking lot.

[0009] Step 2: Before starting the detection, there is no car in the parking lot. The transmitting antenna transmits several sensing signals, and at the same time, reconfigures the RIS phase multiple times according to the requirements of the sensing algorithm. After receiving the sensing signals, the receiving antenna performs channel estimation and extracts the channel components that are scattered by the transmitting antenna through the ROI and RIS in sequence and finally reach the receiving antenna as the reference channel measurement corresponding to the concrete floor of the parking lot.

[0010] Step 3: When starting the parking lot detection, repeat the process of Step 2, extract the channel components that are scattered by the transmitting antenna through the ROI and RIS in sequence and finally reach the receiving antenna as the measurement of the parking lot during detection. Then subtract the reference measurement to capture the possible impact of the presence of vehicles on the channel.

[0011] Step 4: Divide the imaging target, i.e., the parking lot, into several spatial grid units. Each several units can cover a parking space, and the vector composed of the equivalent scattering coefficients of each unit is the solution target of imaging. Generate a sensing matrix based on the known positions of the transmitter and receiver, the position of the RIS, and the grid position.

[0012] Step 5: Using the channel measurement difference obtained in Step 3 and the sensing matrix obtained in Step 4, implement imaging of the parking lot using a compressive sensing algorithm, and determine the occupancy status of the parking spaces based on the imaging results.

[0013] Further, in Step 1, the distributed RIS system collaborates using T blocks of RIS, where the t-th block of RIS contains M t units, and the spatial vector represents the position of the m-th unit of the t-th block of RIS. The T blocks of RIS are respectively installed at different positions Z meters above the ground of the parking lot. The phase configuration of the RIS is regulated by the controller. The RIS array plane is parallel to the ground and faces downward to form an imaging aperture. The number of transmitting antennas installed in the parking lot is N T and the vector represents the placement position of the i-th transmitting antenna; the number of receiving antennas is N R and the vector represents the placement position of the j-th receiving antenna.

[0014] Further, in Step 2, the transmitting antenna transmits the sensing signal K times. When transmitting the sensing signal, the RIS phase is changed once every β symbol times, and the RIS phase is changed N times in total, where K = βN. The RIS phase configuration can be obtained by any one of the methods such as random generation, simulated annealing algorithm, genetic algorithm, and hybrid genetic simulated annealing phase optimization algorithm.

[0015] Further, in Step 2, after the receiving antenna receives the signal, channel estimation is performed, and the channels scattered by different RISs are extracted based on the angle of arrival AOA. Among them, the channel measurement of the sensing signal sent by the i-th transmitting antenna for the k-th time, reflected by the ROI and the t-th block of RIS and finally reaching the j-th receiving antenna is denoted as y 0 i,k,t,j , and this measurement is a scalar. The measurements extracted through the receiving-end channel estimation on all N R receiving antennas are written as a vector Stacking all the measurements corresponding to the transmitting antennas, receiving antennas, RISs, and different transmitted signals, the reference channel measurement corresponding to the concrete floor of the parking lot can be obtained, denoted as where N A = N R × N T × K × T. The estimation of AOA can be obtained by any one of multiple signal classification, MUSIC (multiple signal classification), ESPRIT (estimation of signal parameters via rotational invariance techniques), orthogonal matching pursuit OMP, and deep learning algorithms.

[0016] Further, in step 3, when starting the parking lot detection, repeat step 2, and record the channel measurement of the sensing signal emitted by the i-th transmitting antenna for the k-th time, passing through the ROI and the t-th RIS and finally reaching the j-th receiving antenna as y 1 i,k,t,j , and write the measurements extracted through the receiver channel estimation on all N R receiving antennas as a vector Then stack all y 1 i,k,t to obtain the channel measurement where N A = N R × N T × K × T. Subtract the reference measurement y 1 from y 0 , and denote the difference as y. y captures the impact of possible vehicle presence on the channel and is the measurement for imaging.

[0017] Further, in step 4, divide the imaging target, i.e., the ROI, into Q spatial grid cells, and every C spatial grid cells can cover one parking space. Among them, the spatial vector at the position of the q-th spatial grid cell is denoted as Before starting the detection, the parking lot is empty, and its equivalent scattering coefficient is x 0 q , and the equivalent scattering coefficients of each unit form a column vector of length Q When starting the parking lot detection, its equivalent scattering coefficient is x 1 q , and the equivalent scattering coefficients of each unit form a column vector of length Q Subtract x 1 from x 0 , and denote the obtained difference as Δx where Δx q = x 1 q ― x 0 q . Δx is the target to be solved.

[0018] For the propagation path of the sensing signal from the transmitting antenna, passing through the ROI and the t-th RIS and reaching the receiving antenna, the sensing signal emitted by the i-th transmitting antenna for the k-th time, after passing through the q-th spatial grid cell and the m-th RIS unit of the t-th RIS, finally reaches the j-th receiving antenna. The channel experienced by this sensing signal can be expressed as:

[0019]

[0020] where g is a constant determined by the transmitting and receiving gains of the antenna; λk Denotes the wavelength of the carrier frequency where the k-th sensing signal is located; x q Denotes the scattering coefficient of the q-th spatial grid cell. Before starting the detection, there are no vehicles in the parking lot, x q = x 0 q , when starting the parking lot detection, x q = x 1 q ; Denotes the RIS phase configuration of the m-th unit of the t-th RIS when the k-th sensing signal is transmitted. Summing up all the sub-channels reflected by the Q spatial grid cells and the M units on the t-th RIS, the sum can be expressed as: t

[0021]

[0022] where

[0023]

[0024] where, Denotes the phase configuration vector of the t-th RIS when the i-th transmitting antenna emits the k-th sensing signal; Denotes the sub-channel from the t-th RIS to the j-th receiving antenna; and respectively denote the sub-channels from the i-th transmitting antenna to the ROI and from the ROI to the t-th RIS. These channels can all be calculated based on physical parameters such as the positions of the known transmitting antennas, ROI, RIS, and receiving antennas.

[0025] Denote the sensing matrix corresponding to the signal path of the k-th sensing signal emitted by the i-th transmitting antenna, passing through the ROI and the phase configuration of the t-th RIS as where

[0026]

[0027] The channel measurement difference y obtained in step 3 satisfies:

[0028] y = A×Δx + n

[0029] where, is the additive noise caused by the received signal noise and channel estimation error.

[0030] ​Furthermore, in step 5, the channel difference obtained in step 3 has sparsity. That is, not all areas in the parking lot have vehicles parked, and there are many concrete floor units. Therefore, there are many zero elements in the channel difference obtained in step 3. Although the number of measurements is not sufficient to solve all its exact solutions, based on its sparsity, a compressive sensing algorithm can be used to solve the sparse vector and obtain the equivalent scattering coefficient, thus completing the imaging of the target. The algorithm for solving the sparse vector is any one of the OMP, Subspace pursuit (SP), and Compressive Sampling Matching Pursuit (CoSaMP) algorithms.

[0031] Finally, in step 5, after solving the sparse vector Δx, the occupancy status of the parking space is judged. First, it is judged whether there is a target on the spatial grid unit, and then the status of the parking space is judged through the status of the spatial grid unit. Take τ as the judgment threshold for the status of the spatial grid unit. When the estimated result of the relative scattering coefficient of the m-th unit is greater than τ, that is, when the difference between the actual scattering coefficient of the unit and the scattering coefficient of the concrete floor is greater than τ, it is judged that there is a target in this unit; otherwise, it is judged that there is no target. When there is a target in any spatial grid unit corresponding to the p-th parking space, it is judged that there is a vehicle in this parking space; otherwise, it is judged that there is no vehicle in this parking space.

[0032] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: The present invention utilizes the ability of RIS to achieve sensing through electromagnetic wave reflection, without relying on optical induction, and can avoid the influence of extreme conditions such as rain, snow, haze, and low light at night. In a low-light environment, the detection accuracy of the present invention is higher than that of the camera scheme. Since RIS does not rely on magnetic induction, in areas with dense metal interference, the present invention has a significant advantage over geomagnetic sensors. In addition, RIS is a passive device, without the need for power supply and complex signal processing modules, and the cost of RIS is significantly reduced compared with geomagnetic sensors; RIS has no active emission module, and the overall power consumption of the system is significantly reduced compared with the camera scheme. The present invention also uses multiple RISs to construct a distributed RIS system to assist imaging, and compared with only using a single RIS, the detection accuracy is further improved, and it can be competent for the parking space status detection work. Description of the Drawings

[0033] Figure 1 is a flowchart of a method for detecting the occupancy status of a parking space based on RIS-assisted imaging.

[0034] Figure 2 is a schematic diagram of the scenario of a method for detecting the occupancy status of a parking space based on RIS-assisted imaging in an embodiment of the present invention.

[0035] Figure 3It is a curve of the parking space - level accuracy rate (which means the ratio of the total number of vehicles correctly predicted in multiple tests to the total number of vehicles to be predicted) obtained by using a RIS - assisted imaging - based parking space status detection method in an embodiment of the present invention with respect to the actual parking quantity in the parking lot.

[0036] Figure 4 It is a curve of the total accuracy rate (which means the ratio of the number of tests with completely correct predictions of the entire parking lot status in multiple tests to the total number of tests) obtained by using a RIS - assisted imaging - based parking space status detection method in an embodiment of the present invention with respect to the actual parking quantity in the parking lot. Detailed implementation manners

[0037] The present invention will be specifically explained below in combination with the accompanying drawings and an embodiment of a RIS - assisted imaging - based parking space status detection method:

[0038] I. System model adopted in this embodiment

[0039] The distributed RIS system uses T blocks of RIS for cooperation. The t - th block of RIS contains M t units, and the position of the m - th unit of the t - th block of RIS is represented by the spatial vector . The T blocks of RIS are respectively installed at different positions Z meters above the ground of the parking lot. In this embodiment, the number of RIS is taken as T = 2; the two blocks of RIS contain M1 = M2 = 2500 units, and their scattering coefficients can be independently adjusted; the installation height Z of the two blocks of RIS is 3 meters, and they are symmetrically placed above the horizontal axis of the ROI about the longitudinal axis of the ROI and are called RIS1 and RIS2. RIS1 and RIS2 have the same orientation, size, and number of units. Both RIS1 and RIS2 are parallel to the ROI plane. The number of transmitting antennas installed in the parking lot is N T , and in this embodiment, N T is taken as 1, and the placement position of the i - th transmitting antenna is represented by the vector . In this embodiment, since N T = 1, the vector is written as The number of receiving antennas is N R , and in this embodiment, N R is taken as 1, and the placement position of the j - th receiving antenna is represented by the vector . In this embodiment, since N R = 1, the vector is written as

[0040] The transmitting antenna transmits pilot signals a total of K times. In this embodiment, K = 300. When transmitting sensing signals, the RIS phase is changed once every β symbol times, and the RIS phase is changed a total of N times, where K = βN. In this embodiment, the number of symbol times β used for changing the RIS phase once is taken as 1, so the total number of RIS phase changes N = 300.

[0041] The imaging target, i.e., the ROI, is divided into Q spatial grid cells, and every C spatial grid cells can cover a parking space. In this embodiment, the ROI is a cuboid region whose faces are parallel to the coordinate planes. Take the number of spatial grid cells Q

[0042] = 120, the number of spatial grid cells C required to cover a parking space = 2, and each spatial grid cell has its equivalent scattering coefficient x q , and all the scattering coefficients form a column vector of length 120. The spatial vector at the position of the q-th ROI unit is denoted as The size of the ROI is 25 meters in the horizontal direction, divided into 10 parking spaces, each with a width of 2.5 meters; 30 meters in the vertical direction, with a total of 4 rows of parking spaces, each with a length of 5 meters. There is a 2.5-meter interval between the first and second rows of parking spaces, a 5-meter interval between the second and third rows of parking spaces, and a 5-meter interval between the third and fourth rows of parking spaces, which is the driving lane. The scene is as shown in the appendix Figure 2 shown.

[0043] Before starting the detection, there are no cars in the parking lot, x q = x 0 q , and the equivalent scattering coefficients of each unit form a column vector of length 120 In this embodiment, x 0 satisfies a Gaussian distribution with a mean of 0.5 and a variance of 0.1; when starting the parking lot detection, x q = x 1 q , and the equivalent scattering coefficients of each unit form a column vector of length Q In this embodiment, most of the elements in x 1 are equal to the elements at the corresponding positions in x 0 , and a small number of elements satisfy a random distribution with a lower limit of 0.85 and an upper limit of 0.95. The number of these elements is twice the actual number of parked cars in the parking lot. Subtract x 1 from x 0 , and the obtained difference is denoted as Δx, where, Δx q = x 1 q ―x 0 q . Δx is the target to be solved.

[0044] In this embodiment, for the propagation path of the sensing signal emitted from the transmitting antenna, passing through the ROI and the t-th RIS block and reflected to the receiving antenna, the sensing signal emitted by the transmitting antenna at the k-th transmission, after passing through the q-th spatial grid unit and the m-th RIS unit of the t-th RIS block, finally reaches the receiving antenna. The channel experienced by this sensing signal can be expressed as:

[0045]

[0046] where g is a constant determined by the transmitting and receiving gains of the antennas; λ k represents the wavelength of the carrier frequency where the k-th sensing signal is located; represents the RIS phase configuration of the m-th unit of the t-th RIS block when the k-th sensing signal is transmitted.

[0047] II. Specific Steps of this Embodiment

[0048] The flowchart of the present invention is as shown in the appendix Figure 1 and can be divided into 5 steps, specifically:

[0049] Step (1)

[0050] Set up two RISs to construct a distributed RIS system, and set up a transmitting antenna and a receiving antenna. Physical parameters such as the number of units of the RIS, the installation location of the RIS, the installation locations of the transmitting antenna and the receiving antenna, etc. are described as in the system model of this embodiment above.

[0051] Step (2)

[0052] The transmitting antenna transmits 300 sensing signals in total. When transmitting the sensing signals, the RIS phase is changed once every 1 symbol time, and the RIS phase is changed 300 times in total. In this embodiment, the RIS phase configuration is obtained by random generation. After the receiving antenna receives the signal, channel estimation is performed, and the channels scattered by different RISs are extracted based on the angle of arrival (AOA). Among them, the channel measurement of the sensing signal emitted by the transmitting antenna for the k-th time, passing through the ROI and reflected by the t-th RIS block and finally reaching the receiving antenna, is denoted as y 0 k,t , and this measurement is a scalar. Stacking up the measurements corresponding to all the transmitted signals passing through the two RISs, the reference channel measurement corresponding to the concrete floor of the parking lot can be obtained, denoted as where N A = N R ×N T ×K×T. In this embodiment, N R = N T = 1, K = 300, T = 2, then N A = 600, that is

[0053] Step (3)

[0054] When starting the parking lot detection, repeat Step (2), and denote the channel measurement of the sensing signal sent by the transmitting antenna for the k-th time, passing through the ROI and reflected by the t-th RIS block and finally reaching the receiving antenna as y 1 k,t , and then stack all y 1 k,t to obtain the channel measurement Subtract the reference measurement y 1 from y 0 , and denote the difference as y

[0055] Step (4)

[0056] Stack all the sub-channels reflected by 120 spatial grid cells and 2500 cells on the t-th RIS block, and their sum can be expressed as:

[0057]

[0058] where

[0059]

[0060] Denote the sensing vector corresponding to the signal path of the sensing signal sent by the transmitting antenna for the k-th time, passing through the ROI and the phase configuration of the t-th RIS block as A k,t = h T k,t ∈ C 1×120 , where

[0061]

[0062] The channel measurement difference y obtained in Step (3) satisfies:

[0063] y = A × Δx + n

[0064] where n ∈ C 600×1 is the additive noise caused by the received signal noise and channel estimation error

[0065] Step (5)

[0066] According to the sparsity of Δx, the compressive sensing algorithm can be used to solve the sparse vector and complete the imaging of the target. In this embodiment, the algorithm for solving the sparse vector is the Subspace Pursuit (SP) algorithm. After solving the sparse vector Δx, the occupancy status of the parking space is judged. First, it is judged whether there is a target on the spatial grid unit, and then the status of the parking space is judged through the status of the spatial grid unit. Take τ as the judgment threshold for the status of the spatial grid unit. In this embodiment, the judgment threshold τ = 0.3 is taken. When the estimated result of the relative scattering coefficient of the m-th unit is greater than 0.3, it is judged that there is a target in this unit; otherwise, it is judged that there is no target. When there is a target in any spatial grid unit corresponding to the p-th parking space, it is judged that there is a vehicle in this parking space; otherwise, it is judged that there is no vehicle in this parking space.

[0067] III. Simulation experiment results of this embodiment

[0068] In this invention, Matlab is used as the simulation platform. In order to evaluate the gain of the perception accuracy of this invention, under the condition of different actual parking numbers in the parking lot, based on this example, two evaluation criteria, namely the parking space-level accuracy (the ratio of the total number of correctly predicted vehicles to the total number of vehicles to be predicted under multiple tests) and the overall accuracy (the ratio of the number of tests with completely correct predictions of the entire parking lot state to the total number of tests under multiple tests), are introduced, and the curves of these two values with respect to the actual parking number in the parking lot are given. The simulation results are shown in Appendix Figure 3 、Appendix Figure 4 respectively. According to the analysis of this simulation result, for a parking lot with 40 parking spaces and a size of 25 meters * 30 meters, when the actual number of parked vehicles in the parking lot is less than 8, both the parking space-level accuracy and the overall accuracy generally fluctuate and rise with the increase of the actual number of parked vehicles in the parking lot, and always maintain a relatively high accuracy of more than 80%. When the actual number of parked vehicles in the parking lot reaches 10, both indicators reach 100%, and until the actual number of parked vehicles in the parking lot reaches 28, both indicators have been or approximately reached 100%. After the actual number of parked vehicles in the parking lot exceeds 28, both the parking space-level accuracy and the overall accuracy decrease significantly with the increase of the actual number of parked vehicles in the parking lot. In particular, the overall accuracy is always 0, indicating that the actual number of vehicles in the parking lot has reached a relatively large scale at this time, making the vector Δx to be solved no longer sparse, and the performance of the compressive sensing algorithm has dropped significantly.

[0069] This simulation result confirms that a method for detecting the status of parking spaces based on RIS-assisted imaging provided by this invention has a quite high detection accuracy when the actual number of vehicles in the parking lot is within a certain range, and has practical application capabilities. And this method has two characteristics compared with the original methods for detecting the status of parking spaces on the market, namely low deployment economic cost and low facility layout complexity, and has market application value.

[0070] The above combines the accompanying drawings to provide a detailed introduction to a method for detecting the status of a parking space based on RIS-assisted imaging. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are illustrative and not restrictive. The embodiments are only used to illustrate the technical idea of the present invention and do not limit the protection scope of the present invention. Those of ordinary skill in the art will make changes in the specific embodiments and application scope under the inspiration of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting the status of a parking space based on RIS-assisted imaging, characterized in that, Utilize multiple reconfigurable intelligent surfaces (RISs) to collaboratively complete two-dimensional imaging of the target region (ROI), i.e., the parking lot ground, and then achieve monitoring of the parking space status. The method includes the following steps: Step 1: Install multiple RISs at different positions several meters above the parking lot ground to construct a distributed RIS system, and install a transmitting antenna and a receiving antenna at appropriate positions in the parking lot. Step 2: Before starting the detection, there are no cars in the parking lot. The transmitting antenna transmits several sensing signals, and at the same time, reconfigures the RIS phase multiple times according to the requirements of the sensing algorithm. After the receiving antenna receives the sensing signals, it performs channel estimation and extracts the channel components that are scattered by the transmitting antenna through the ROI and RISs in sequence and finally reach the receiving antenna as the reference channel measurement corresponding to the parking lot concrete ground. Step 3: When starting the parking lot detection, repeat the process of Step 2, extract the channel components that are scattered by the transmitting antenna through the ROI and RISs in sequence and finally reach the receiving antenna as the measurement of the parking lot during detection, and then subtract the reference measurement to capture the impact on the channel caused by the possible presence of vehicles. Step 4: Divide the imaging target, i.e., the parking lot, into several spatial grid units, with each several units covering one parking space. The vector composed of the equivalent scattering coefficients of each unit is the solution target of the imaging. Generate a sensing matrix based on the known positions of the transmitter, receiver, RISs, and grid positions. Step 5: Use the channel measurement difference obtained in Step 3 and the sensing matrix obtained in Step 4, and use the compressive sensing algorithm to achieve imaging of the parking lot, and judge the occupancy status of the parking space based on the imaging result.

2. The method for detecting the state of a parking space based on RIS-assisted imaging according to claim 1, wherein, In Step 1, the distributed RIS system collaborates using T RIS blocks. The t-th RIS block contains M t units, and the position of the m-th unit of the t-th RIS block is represented by the spatial vector . The T RIS blocks are respectively installed at different positions Z meters above the parking lot ground. The phase configuration of the RIS is regulated by the controller, and the RIS front is parallel to the ground and faces downward to form an imaging aperture.

3. The method for detecting the state of a parking space based on RIS-assisted imaging according to claim 1, wherein, In step 1, the number of transmitting antennas installed in the parking lot is N T , and the vector represents the placement position of the i-th transmitting antenna; the number of receiving antennas is N R , and the vector represents the placement position of the j-th transmitting antenna.

4. A method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, wherein In Step 2, the transmitting antenna transmits a total of K sensing signals. When transmitting the sensing signals, the RIS phase is changed once every β symbol times, and the RIS phase is changed a total of N times, where K = βN. The RIS phase configuration can be obtained through any one of methods such as random generation, simulated annealing algorithm, genetic algorithm, and hybrid genetic simulated annealing phase optimization algorithm.

5. The method for detecting the parking space status based on RIS-assisted imaging according to claim 1, wherein In step 2, after the receiving antenna receives the signal, channel estimation is performed, and the channels scattered by different RISs are extracted based on the angle of arrival (AOA). Among them, the channel measurement of the sensing signal sent by the i-th transmitting antenna for the k-th time, after being reflected by the ROI and the t-th RIS block and finally reaching the j-th receiving antenna, is denoted as y 0 i,k,t,j , this measurement is a scalar, and the measurements extracted through receiver channel estimation on all N R receiving antennas are written as a vector Stacking the measurements corresponding to all transmitting antennas, receiving antennas, RISs, and different transmitting signals, the reference channel measurement corresponding to the concrete floor of the parking lot can be obtained, denoted as where N A = N R × N T × K × T, and the estimation of AOA is obtained through any one of multiple signal classification (MUSIC), estimation of signal parameters via rotational invariance techniques (ESPRIT), orthogonal matching pursuit (OMP), and deep learning-based algorithms 6. The method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, wherein, In step 3, when the parking lot detection starts, step 2 is repeated, and the channel measurement of the k-th sensing signal sent by the i-th transmitting antenna, which is reflected by the ROI and the t-th RIS and finally reaches the j-th receiving antenna is recorded as y 1 i,k,t,j , all N R The measurement extracted by the receiver channel estimation at the root receiving antenna is written as the vector Then all y 1 i,k,t Stack them together to get channel measurements Where N A =N R ×N T ×K×T, use y 1 Subtract the reference measurement y 0 , the difference is recorded as y, which captures the impact of possible vehicle presence on the channel and is a measurement used for imaging.

7. A method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, characterized in that, In step 4, the imaging target, i.e., the ROI, is divided into Q spatial grid cells, and every C spatial grid cells cover one parking space. Among them, the spatial vector at the position of the q-th spatial grid cell is denoted as Before starting the detection, there is no vehicle in the parking lot, and its equivalent scattering coefficient is x 0 q , and the equivalent scattering coefficients of each cell form a column vector with a length of Q When starting the parking lot detection, its equivalent scattering coefficient is x 1 q , and the equivalent scattering coefficients of each cell form a column vector with a length of Q Use x 1 Subtract x 0 , and the obtained difference is denoted as Δx, Among them, Δx q = x 1 q ― x 0 q , and Δx is the target to be solved 8. A method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, wherein, In Step 4, for the propagation path of the sensing signal emitted from the transmitting antenna, reflected by the ROI and the t-th RIS, and reaching the receiving antenna, for the sensing signal emitted by the i-th transmitting antenna at the k-th transmission, after passing through the q-th spatial grid unit and the m-th RIS unit of the t-th RIS, and finally reaching the j-th receiving antenna, the channel experienced by this sensing signal is expressed as: where g is a constant determined by the transmitting and receiving gains of the antenna; λ k represents the wavelength of the carrier frequency where the k-th sensing signal is located; x q represents the scattering coefficient of the q-th spatial grid cell. Before starting the detection, there are no vehicles in the parking lot, and x q = x 0 q . When starting the parking lot detection, x q = x 1 q ; represents the RIS phase configuration of the m-th unit of the t-th RIS when the k-th sensing signal is transmitted. Adding up all the sub-channels reflected by the Q-th spatial grid cell and the M t units on the t-th RIS, the sum can be expressed as: where Among them, represents the phase configuration vector of the $t$-th RIS when the $i$-th transmit antenna emits a sensing signal for the $k$-th time; represents from the sub-channel of the $t$-th RIS to the $j$-th receive antenna; and respectively represent the sub-channels from the $i$-th transmit antenna to the ROI and from the ROI to the $t$-th RIS, and these channels are all calculated based on physical parameters such as the positions of the known transmit antennas, ROI, RIS, and receive antennas. Denote the sensing matrix corresponding to the signal path of the sensing signal transmitted by the \(i\)-th transmitting antenna for the \(k\)-th time, passing through the ROI and the phase configuration of the \(t\)-th RIS as where The channel measurement difference y obtained in Step 3 satisfies: y = A×Δx + n wherein, is the additive noise caused by the received signal noise and the channel estimation error.

9. The method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, wherein, In Step 5, based on the compressive sensing theory, solve the sparse vector Δx, and this solving process is completed using algorithms such as OMP (Orthogonal Matching Pursuit), Subspace Pursuit (SP), and Compressive Sampling Matching Pursuit (CoSaMP).

10. A method for detecting the status of a parking space based on RIS-assisted imaging according to claim 1, characterized in that, In Step 5, after solving the sparse vector Δx, judge the occupancy status of the parking space. First, judge whether there is a target on the spatial grid unit, and then judge the status of the parking space through the status of the spatial grid unit. Take τ as the judgment threshold for the status of the spatial grid unit. When the estimated result of the relative scattering coefficient of the m-th unit is greater than τ, that is, when the difference between the actual scattering coefficient of this unit and the scattering coefficient of the concrete ground is greater than τ, it is judged that there is a target in this unit. Otherwise, it is determined that there is no target. When there is a target in any of the spatial grid cells corresponding to the p-th parking space, it is determined that there is a vehicle in this parking space; otherwise, it is determined that there is no vehicle in this parking space.

Images