A multi-RIS-assisted location information perception method based on WIFI fingerprint

Abstract

The present invention discloses a location information perception method based on WIFI fingerprint assisted by multiple RIS, comprising: a WIFI terminal at an unconventional location k first receives a reference signal with relatively low signal quality from a certain AP and senses whether the non-direct path transmission condition is met. The AP is responsible for controlling the working order of multiple RIS and the reflection phase at the RIS, and the WIFI terminal then sequentially receives the AP reference signal forwarded through two or more reconfigurable smart surface RIS, so as to maximize the receiving strength value of the WIFI terminal and avoid interference from the RIS. The WIFI device converts the AP reference signal and the reference signals from the two RIS into a two-dimensional feature matrix, so as to perform classification algorithms such as low-complexity variable decision trees at the terminal. Even in a single AP configuration and non-line-of-sight environment, the spatial resolution of a specific sub-area is improved through multiple RIS and machine learning algorithms, thereby supplementing the ranging-based wireless positioning service.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor Wi-Fi perception and monitoring, and in particular to a multi-RIS-assisted Wi-Fi fingerprint-based location information perception method. Background Art

[0002] With the development of wireless LAN technology, Wi-Fi has overcome the limitations of connecting devices via network cables, eliminating the need for complex wired interface technology and instead using radio waves for networking. The current explosive growth of mobile devices, smart wearables, and other terminals has made Wi-Fi technology a necessity in every household, and Wi-Fi signals are virtually ubiquitous. Therefore, in the process of integrating perception and communication, Wi-Fi signals with perception capabilities have a wide range of applications.

[0003] However, due to the broadcast nature of Wi-Fi wireless signals, user communications face security threats such as interference, blocking, and eavesdropping. Interference has become a significant factor in degrading signal reception quality. Most existing interference management methods assume that the expected interference is known to the communication device and directly manage the known interference. In practice, however, it is uncertain whether the receiver is experiencing interference. Therefore, before performing interference management, it is necessary to first determine whether interference exists in the environment and then, based on the judgment result, determine whether interference management is necessary to avoid increased communication resource overhead, complexity, and even performance loss caused by blind interference management. Furthermore, even if communication devices are expected to attempt to detect some state information about the interference source, on the one hand, the measurement will incur significant overhead, and on the other hand, not all state information is measurable, which will hinder further interference management. Therefore, detecting environmental interference and perceiving interference characteristics independently of cooperation are crucial for effective interference management, yet they are also challenging. Summary of the Invention

[0004] Purpose of the Invention: To address the above issues, this paper proposes a multi-RIS-assisted Wi-Fi fingerprint-based location information sensing method. By applying Reconfigurable Intelligent Surface (RIS) technology to indoor Wi-Fi sensing and monitoring, this method not only enables interference monitoring during the positioning and sensing process, but also improves the positioning accuracy, robustness, and cross-scenario capabilities of the Wi-Fi sensing system.

[0005] Technical solution: To achieve the purpose of the present invention, the technical solution adopted by the present invention is: a multi-RIS-assisted location information perception method based on WIFI fingerprint, comprising the following steps:

[0006] Step 1: Deploy a sensing monitoring system indoors consisting of a RIS device, a Wi-Fi signal transmitter, and a Wi-Fi signal receiver. The RIS device is composed of multiple reflective units, which are evenly laid out in two dimensions on the RIS plane and connected to a computer control terminal via wires.

[0007] Step 2: Determine whether there is blockage or interference on the LOS path in the sensing monitoring system based on the received signal strength at the receiving end. If so, execute step 3; otherwise, execute step 5.

[0008] Step 3: Enable the RIS device, establish a virtual LOS path without obstruction, and generate a virtual Fresnel zone model for perception;

[0009] Step 4: Process the RSS-containing signals collected by the receiver and use a machine learning algorithm to train and test the perception monitoring system that generates the Fresnel zone model. Adjust the position and angle of the RIS to maximize the signal-to-noise ratio at the receiver. Once the accuracy requirements are met, the perception training ends.

[0010] Step 5: Maintain the state of the perception monitoring system after training and testing and adjustment by the RIS device, and use the Wi-Fi positioning algorithm to obtain the positioning coordinates of the user end;

[0011] Step 6: If the signal-to-noise ratio of the received signal at the receiving end is continuously below the threshold, it is determined that the virtual LOS path is blocked, and the process returns to step 3.

[0012] Furthermore, the method for determining whether there is obstruction or interference on the LOS path in step 2 is:

[0013] By the transmitter T x By designing the reflection coefficient matrix of the RIS intelligent reflector, the expected signal received by the receiver is zero, and the interference at the receiver is found, including:

[0014] (1) The receiver estimates the channel state information between it, the transmitter, and the RIS and feeds it back to the transmitter;

[0015] (2) The transmitter adjusts the reflection coefficients of the K reflection units on the RIS to Constructing the reflection matrix Make the expected signal received by the receiving end zero; where β and θ0 are the adjustable amplitude and phase of the reflection unit, j is a complex unit, E K is a unit matrix with an order of K; if the signal level received by the receiving end exceeds a preset decision threshold, it is determined that the receiving end is interfered with.

[0016] Furthermore, when the number of wireless access points (APs) is less than 3 and the number of RISs is greater than 1, the Wi-Fi positioning algorithm uses a training set to train a variable decision tree or a low-complexity machine learning algorithm offline or online.

[0017] Furthermore, the Wi-Fi positioning algorithm divides the Wi-Fi signal coverage area into grids and sub-grids, and labels the grids and sub-grids separately to establish an effective feature space;

[0018] Under unconventional wireless propagation conditions, that is, when each AP has no direct path to the secondary subgrid (sub-area), multiple RISs are used to sense location information in the unconventional area.

[0019] Furthermore, the Wi-Fi positioning algorithm uses a decision tree or a low-complexity machine learning classification algorithm at the receiving end to adjust the reflection phase of the RIS, convert the AP's reference signal and the reference signals from the two RIS into a two-dimensional feature matrix covering the sub-grid areas divided by the Wi-Fi signal, output sub-area labels, and feed them back to the system receiving end.

[0020] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0021] (1) The method of the present invention introduces RIS technology into indoor Wi-Fi perception monitoring, such as Figure 1 As shown in the figure, when the LOS path between the transmitter and receiver is blocked by an obstacle, a virtual line of sight (LOS) path can be generated through RIS, which improves the cross-scenario capability of the Wi-Fi perception monitoring system while ensuring perception accuracy.

[0022] (2) In the method of the present invention, RIS can be used to determine whether there is obstruction or interference in the line-of-sight path, thereby reducing the cost and complexity of communication resources.

[0023] (3) The RIS in the method of the present invention can be adjusted in position and angle. When obstacles or interferences exist in the configured perception monitoring system, the perception function can be restored by simply adjusting the RIS, thereby improving the robustness of the Wi-Fi perception monitoring system.

[0024] (4) The method of the present invention applies RIS to the Wi-Fi signal sensing solution. The low-cost, low-power RIS will greatly improve the performance of the system without changing the existing Wi-Fi equipment, achieving low-complexity and cost-effective gain improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a schematic diagram of a model of a display scene used in the method of the present invention;

[0026] Figure 2 is a flow chart of the method of the present invention;

[0027] Figure 3 Schematic diagram of a model for determining whether there is obstruction interference on the LOS path according to the present invention;

[0028] Figure 4 This is an algorithm flow model diagram of the fingerprint positioning method based on machine learning;

[0029] Figure 5 It is a topological diagram of the regional awareness system assisted by multiple RIS;

[0030] Figure 6 It is a classification tree with a pruning level of 5 for sub-region classification;

[0031] Figure 7 The most important predictor of the variable decision tree algorithm is X2;

[0032] Figure 8 is the received signal distribution from RIS1 at the jth sub-region in position k;

[0033] Figure 9 is the received signal distribution from RIS2 at the j-th sub-region in position k. DETAILED DESCRIPTION

[0034] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0035] The present invention implements the RIS-assisted Wi-Fi perception monitoring method provided by Figure 2 As shown, the steps are as follows:

[0036] Step 1: Deploy a sensing monitoring system indoors consisting of multiple RIS, Wi-Fi signal transmitters, and Wi-Fi signal receivers. The RIS device consists of multiple reflective units, which are evenly laid out in two dimensions on the RIS plane and connected to a computer control terminal via wires.

[0037] The RIS device, taking into account the established wireless environment configuration, contains N R There are three intelligent reflective surface devices, each containing N antenna units. The channel coefficients of the three links are defined as follows: the channel coefficient h from the Ath wireless access point (AP) to the target point k (link is denoted as Ak) A,k , the channel coefficient h from the Ath AP to the Rth smart reflective surface device (link is represented as AR) A,R,i , the channel coefficient h from the Rth smart reflective surface device to position k (link is denoted as Rk) R,i,k, which are expressions (1), (2) and (3) respectively:

[0038]

[0039]

[0040]

[0041] Where A={1,…,N A}, R={1,…,N R}, i={1,…,N}, k={1,…,N k};N A 、N R 、N k are the number of wireless access nodes, smart reflective surface devices and target point locations, respectively, j is a complex unit; a A,k and θ A,k They represent the amplitude and phase of the channel coefficient of link Ak, respectively. A,R,i and θ A,R,i Respectively represent the amplitude and phase of the channel coefficient of link AR, a R,i,k and θ R,i,k Represent the amplitude and phase of the channel coefficient of link Rk respectively.

[0042] Define the transmission signals of all APs as s, w k represents the background noise at position k, and the phase of the i-th unit in the R-th device is θ R,i ,θ R,i ={0,π / 2,π,3π / 2}.

[0043] After the single-hop link (path) signal and the two-hop link (path) signal (using only the Rth smart reflective surface device) are superimposed at the terminal, the total received signal is

[0044]

[0045] If the first term and the second term on the right side of equation (4) have the same phase, the received power increases, thereby reducing the impact of shadow fading. This is one of the ideas of the method of the present invention.

[0046] Note that if the AP employs multiple transmit antennas and the terminal employs multiple receive antennas, the total received signal from the transmitter (Tx) to the receiver (Rx) will be the vector sum of the multiple paths.

[0047] Step 2: Determine whether there is any obstruction or interference on the LOS path in the system based on the received signal strength at the receiving end. If so, proceed to step 3; otherwise, proceed to step 5.

[0048] The method to determine whether there is blocking or interference is as follows: x By designing the reflection coefficient matrix of the RIS intelligent reflector, the expected signal received by the receiver is made zero, thereby finding the interference at the receiver. Figure 3 As shown, the method includes:

[0049] (1) The receiver estimates the channel state information between it, the transmitter, and the RIS and feeds it back to the transmitter;

[0050] (2) The transmitter adjusts the reflection coefficients of the K reflection units on the RIS to Constructing the reflection matrix Make the expected signal received by the receiving end zero; where β and θ0 are the adjustable amplitude and phase of the reflection unit, j is a complex unit, E K is a unit matrix with an order of K; if the signal level received by the receiving end exceeds a preset decision threshold, it is determined that the desired receiving end is interfered with.

[0051] Step 3: Enable one or more RISs to establish a virtual LOS path without obstruction and generate a virtual Fresnel zone model for perception.

[0052] Step 4: Process the received signal strength (RSS) signal collected by the receiver and use a machine learning algorithm (such as a DNN algorithm) to train and test the perception monitoring system that generates the Fresnel zone model. Adjust the position and angle of the RIS to maximize the signal-to-noise ratio at the receiver. Once the accuracy requirements are met, the perception training ends.

[0053] Step 5: Maintain the current state of the perception monitoring system (the state after training, testing, and adjustment by the RIS device) and use a Wi-Fi positioning algorithm (such as fingerprint positioning) to obtain the precise positioning coordinates of the user end.

[0054] Step 6: If the signal-to-noise ratio of the received signal at the receiving end is continuously below 10 dB (ie, the signal is at low strength for a long time), it is determined that the virtual LOS path is blocked, and the process returns to step 3.

[0055] The Wi-Fi positioning algorithm described in step 5 depends on the number of APs in the system. When the number of APs is greater than or equal to 3, a three-sided positioning calculation method based on TOA or RSS can be used; when the number of APs may be less than 3, a fingerprint positioning algorithm based on machine learning is used. Regarding the Wi-Fi positioning algorithm, the embodiment of the present invention proposes three calculation methods. Among them, only the three-sided positioning algorithm based on TOA or RSS used when the number of APs is greater than 3 is used as a control in the simulation. The present invention will focus on the fingerprint positioning algorithm based on machine learning. The algorithm flow is as follows: Figure 4 shown.

[0056] Traditional machine learning classification algorithms include decision trees, support vector machines, nearest neighbor algorithms, and Bayesian networks. They also include clustering algorithms, association analysis algorithms, and regression analysis algorithms. The present invention uses decision trees primarily due to their low implementation complexity and power consumption at the terminal, as well as the potential for online training and testing. If the terminal has high power consumption and computing power, deep learning algorithms can also be used, but this comes with the added complexity of online training. The advantages of decision tree algorithms are summarized below. First, decision trees are easy to understand and implement. Second, they are fast, require minimal computation, and can be easily converted into classification rules. When traversing from the root node down to the leaf nodes, the splitting conditions along the way are unique and deterministic. The main disadvantage of decision tree algorithms is that they are prone to overfitting when processing large sample sets, reducing classification accuracy. Pruning is the primary method used by decision tree algorithms to combat overfitting. Excessive branches may perfectly fit the training data, but not necessarily the test data, and can be inefficient. Therefore, some branches can be proactively removed to reduce the risk of overfitting. Pruning includes pre-pruning and post-pruning, which correspond to pruning during the decision tree generation process / after the decision tree generation is completed.

[0057] Figure 5 This is a topology diagram of multiple RIS-assisted regional awareness systems. Its unconventional wireless propagation conditions mean that each AP has no direct path to reach the sub-area M22. Therefore, in the proposed RIS-assisted location information awareness algorithm (abbreviated as the proposed awareness algorithm), the Wi-Fi coverage area is divided into grids and sub-grids. Figure 5 Here, M1, M2, M3, and M4 represent the first-level grid, and M21, M22, M23, and M24 represent the second-level grid, also known as a sub-area. This definition is not general; M22 is defined here to meet unconventional wireless propagation conditions. Location information in this sub-area can only be detected with the help of multiple RISs. A terminal at M22 will receive reflected signals from both RIS1 and RIS2.

[0058] Scheduling considerations in data collection are briefly described below. Figure 5In this example, AP1 controls two RIS devices, RIS1 and RIS2. To reduce signal interference between the two RIS devices, the proposed sensing algorithm uses a simple protocol. The terminal receiver first receives the signal from the AP, then sequentially receives signals from the two RIS devices. To combat path loss and multiplicative factors in the wireless channel, the two RIS devices are typically located between the AP and the terminal. The number of RIS elements, N, is defined, and the N1 RIS elements are grouped, with the reflection phase within each group being the same. This simple scheduling strategy reduces phase polling time. In the simulation and verification of the present invention, a 2-bit discrete phase is used, followed by combining subvectors of length 4, with each element of the vector selecting a 2-bit discrete phase. There are 24 possible subvector combinations. For example, for a RIS configuration with N = 64 elements, it is divided into 16 groups. Therefore, the present invention proposes a low-complexity reflection phase scheme. This significantly reduces the complexity of the search algorithm for finding the maximum RSSI value at the terminal.

[0059] The Wi-Fi positioning algorithm obtains the user's terminal location coordinates based on the signal's time of arrival (TOA) using a range circle method. The process is as follows:

[0060] Assume that the coordinates of the three transmitters involved in positioning, namely the three anchor points, are recorded as (x i ,y i ), 1≤i≤3, the three anchor points are not on a straight line; the coordinates of the unknown node are (x, y), then the distance measurement equation is expressed as follows:

[0061]

[0062] The above formula is the expression of the distance circle. The target unknown node is located at (x i ,y i ) is the center of the circle, and the distance d i The common area formed by three distance circles with a radius of , is selected as the coordinate of the unknown node at the center of mass point M of the common area.

[0063] To solve the coordinates of the centroid of the common area of ​​the distance circle, the following nonlinear equations are listed:

[0064]

[0065] The measured distance between the unknown node and the anchor point is the input of the nonlinear equation system in Equation (6), and the coordinates of the unknown node are the output. Suppose the coordinates of the three intersection points on the inner side of the distance circle obtained by the nonlinear equation system are (x a ,y a ),(x b ,y b ),(x c ,y c), then the coordinates of the center of mass M are:

[0066]

[0067] This is the location coordinate of the unknown node.

[0068] Furthermore, the Wi-Fi positioning algorithm obtains the positioning coordinates of the user terminal based on the received signal strength RSS, and the process is as follows:

[0069] The received signal power P r (d) has the following relationship with the distance d between the target user and the base station:

[0070]

[0071] Among them, P t Represents the transmission power of the transmitter, i.e. the anchor node, G t Represents its antenna gain; G r represents the antenna gain of the receiving end; λ represents the wavelength of the radio wave; d represents the distance between the transmitting end and the receiving end; L represents the loss factor.

[0072] The free space path loss is expressed as:

[0073]

[0074] Select the near-ground distance d0 as the reference distance for the receiving power of the receiver. The reference distance must be greater than the far-field distance d f , that is, d0≥d f , and must be less than the actual distance between the transmitter and the receiver. When the distance between the transmitter and the receiver is greater than d0, the received power of the receiver in free space is:

[0075]

[0076] The distance between the target and multiple transmitters is calculated based on the signal strength, and then each transmitter is used as the intersection of a circle to represent the location of the target user.

[0077] Furthermore, when the number of APs may be less than 3, a fingerprint positioning algorithm based on machine learning is adopted. In the present invention, a decision tree classification algorithm is selected. The complete processing flow and description of fingerprint-based location information perception are as follows. The following simulation and verification will follow this process.

[0078] The specific process includes:

[0079] 1) The location sensing area is divided into uniform grids (the number of first-level grids is M k =m 2 ) and sub-grids, using mesh functions to generate scene data and visualize them.

[0080] The test position in the traditional three-sided positioning process is modified here, that is, the test position in the traditional positioning process is all grids in the positioning area, but the present invention is aimed at area perception under unconventional wireless propagation conditions, that is, sub-area classification.

[0081] 2) Generate the channel coefficients of the RIS link, record the measurement values ​​from one RIS in sequence under a simple scheduling strategy (such as first-come, first-served strategy), calculate the maximum received signal power value, and pre-process the signal measurement values ​​to obtain the characteristic matrix.

[0082] Here, the terminal uses a search algorithm to find the maximum received signal power value. The Wi-Fi device receiver converts the AP's reference signal and the reference signals from the two RISs into a two-dimensional feature matrix. Preprocessing requires removing outliers such as excessive values ​​(such as 40dB).

[0083] 3) The feature matrix will facilitate the next generation of Wi-Fi systems to monitor whether there is a direct path to the area of ​​interest.

[0084] When a Wi-Fi system retains historical RSSI data, by matching it with newly collected data, it can understand changes in the wireless propagation environment, such as those caused by the repositioning of indoor or outdoor furniture. If a direct path condition exists, a traditional fingerprint wireless positioning algorithm will be run to obtain location information.

[0085] 4) Use public data or independently measure and annotate the collected signal strength data in the test environment to obtain a training set, and use the training set to train variable decision trees or low-complexity machine learning algorithms (such as K-nearest neighbor method, linear regression algorithm, etc.).

[0086] Here we define the label of the j-th sub-region in position k as M kj ,definition Figure 5 Where k = 2, j = {1, 2, 3, 4}. After defining the AP number A, the received power of the j sub-areas from AP1 to position k is defined as RSSI A,k,j =RSSI 1,2,j , the received power from the Rth RIS is defined as RSSI A,R,k,j . Figure 5 Medium M kj The feature space corresponding to the sub-region is defined as [RSSI 1,2,j ,RSSI 1,1,2,j ,RSSI 1,2,2,j ,M kj ];

[0087] During the test, the received power level (fingerprint) is input, and the decision result will output the j-th sub-area label in position k and feed it back to the Wi-Fi location awareness system.

[0088] "X1", "X2" and "X3" are defined to represent the received signal powers from the AP, from RIS1 and from RIS2 respectively. Figure 6 It is a classification tree with a pruning level of 5 for sub-region classification. Figure 7 The most important predictor of the decision tree algorithm is X2, followed by X3. Figure 6 It can be seen that the RIS deployment location has a significant impact on location information perception. Note that in the preprocessing stage, the limitation and calibration of the received signal power are necessary, which affects the performance of the variable decision tree algorithm. Figure 7 It is a classification tree with a pruning level of 5 for sub-region classification, and the number of categories is Figure 1 The actual location-aware solution needs to balance performance and implementation complexity.

[0089] Figure 8 The distribution of received signals from RIS1 at the jth sub-region in position k, Figure 9 Distribution of received signals from RIS2 at the j-th sub-region in position k. Figure 8 and Figure 9 In the table, “direct” and “RIS” represent the received signal power from the AP and from the RIS, respectively. Figures 6 to 9 This strongly supports the feasibility of this plan.

[0090] To illustrate traditional fingerprint recognition wireless positioning solutions, simulation results are presented below. These wireless positioning solutions include trilateration (range circle), multilateration (least squares), batch gradient descent, and stochastic gradient descent. These algorithms employ distance-based approaches, using received power to determine two-dimensional position. For 100 sampling points, the runtimes required for the three traditional methods are 1.039164 seconds, 0.000996 seconds, 0.004590 seconds, and 0.006855 seconds, respectively. The corresponding normalized mean square errors of the coordinate estimates are 1.7716, 0.7825, 0.3926, and 0.3846, respectively. Furthermore, the multilateration (least squares) algorithm is also used. As can be seen from the above, the batch gradient descent algorithm outperforms the trilateration algorithm.

[0091] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction. To avoid unnecessary repetition, the present invention will not further describe various possible combinations.

Claims

1. A multi-RIS-assisted location information perception method based on WIFI fingerprint, characterized in that: The method comprises the following steps: Step 1: Deploy a sensing monitoring system indoors consisting of a RIS device, a Wi-Fi signal transmitter, and a Wi-Fi signal receiver. The RIS device is composed of multiple reflective units, which are evenly laid out in two dimensions on the RIS plane and connected to a computer control terminal via wires. The RIS device contains N R There are smart reflective surface devices, each device contains N antenna units, and defines the channel coefficients of three links: the link from the Ath wireless access node AP to the target point position k is represented as Ak, and the channel coefficient of link Ak is h A,k , the link from the Ath AP to the Rth smart reflective surface device is represented as AR, and the channel coefficient h of link AR is A,R,i , the link from the Rth smart reflective surface device to position k is denoted as Rk, and the channel coefficient h of link Rk is R,i,k , respectively expressed as: Where A={1,…,N A }, R={1,…,N R }, i={1,…,N}, k={1,…,N k }, N A 、N R 、N k are the number of wireless access nodes, smart reflective surface devices and target point locations, respectively, j is a complex unit; a A,k and θ A,k They represent the amplitude and phase of the channel coefficient of link Ak, respectively. A,R,i and θ A,R,i Respectively represent the amplitude and phase of the channel coefficient of link AR, a R,i,k and θ R,i,k Respectively represent the amplitude and phase of the channel coefficient of link Rk; Define the transmission signals of all APs as s, w k represents the background noise at position k, and the phase of the i-th unit in the R-th device is θ R,i ,θ R,i ={0,π / 2,π,3π / 2}; After the single-hop link signal and the two-hop link signal are superimposed at the terminal, the total received signal is The first term on the right side of equation (4) has the same phase as the second term. If the AP uses multiple transmit antennas and the terminal uses multiple receive antennas, the total received signal from the transmitter to the receiver is the vector sum of multiple paths. Step 2: Determine whether there is blockage or interference on the LOS path in the sensing monitoring system based on the received signal strength at the receiving end. If so, execute step 3; otherwise, execute step 5. The method to determine whether there is blocking or interference is as follows: x By designing the reflection coefficient matrix of the RIS intelligent reflector, the expected signal received by the receiver is zero, thereby finding the interference at the receiver. Specifically, the following steps are involved: (1) The receiver estimates the channel state information between it, the transmitter, and the RIS and feeds it back to the transmitter; (2) The transmitter adjusts the reflection coefficients of the K reflection units on the RIS to Constructing the reflection matrix Make the expected signal received by the receiving end zero; where β and θ0 are the adjustable amplitude and phase of the reflection unit, j is a complex unit, E K is a unit matrix of order K; if the signal level received by the receiving end exceeds the preset decision threshold, it is determined that the desired receiving end is interfered with; Step 3: Enable the RIS device, establish a virtual LOS path without obstruction, and generate a virtual Fresnel zone model for perception; Step 4: Process the RSS-containing signals collected by the receiver and use a machine learning algorithm to train and test the perception monitoring system that generates the Fresnel zone model. Adjust the position and angle of the RIS to maximize the signal-to-noise ratio at the receiver. Once the accuracy requirements are met, the perception training ends. Step 5: Maintain the state of the perception monitoring system after training and testing and adjustment by the RIS device, and use the Wi-Fi positioning algorithm to obtain the positioning coordinates of the user end; The Wi-Fi positioning algorithm uses a three-sided positioning calculation method based on the number of APs in the system. When the number of APs is greater than or equal to 3, it uses a TOA or RSS-based three-sided positioning calculation method. When the number of APs is less than 3 and the number of RIS is greater than 1, it uses a fingerprint positioning algorithm based on machine learning. Using the decision tree classification algorithm, the fingerprint-based location information perception processing process is as follows: The location-aware area, i.e., the Wi-Fi signal coverage area, is divided into uniform grids and subgrids. The mesh function is used to generate scene data and visualize it. An effective feature space is established for the grids and subgrids. Generate the channel coefficients of the RIS link, record the measurement values ​​from one RIS in turn under the scheduling strategy, calculate the maximum received signal power value, pre-process the signal measurement values, adjust the reflection phase of the RIS, and the WiFi receiver converts the AP's reference signal and the reference signals from the two RIS into a two-dimensional feature matrix; The feature matrix is ​​used by the next-generation Wi-Fi system to monitor whether a direct path exists in the area of ​​interest. If a direct path exists, under unconventional wireless propagation conditions, where no AP has a direct path to the secondary subgrid, multiple RISs are used to sense the location information in the unconventional area and run a fingerprint wireless positioning algorithm to obtain the location information. The training set is used to train the variable decision tree. The received power level (fingerprint) is input. The decision result outputs the sub-region label of the location and is fed back to the Wi-Fi location awareness system. Step 6: If the signal-to-noise ratio of the received signal at the receiving end is continuously below the threshold, it is determined that the virtual LOS path is blocked, and the process returns to step 3.

Images