RIS-assisted millimeter wave MIMO-OFDM system-based common inductance fusion method

Abstract

The invention discloses a RIS-assisted millimeter wave MIMO-OFDM system-based communication fusion method, which comprises the following steps of: positioning a user and a scatterer, calculating the position of the user by adopting a non-line-of-sight component of an intelligent reflecting surface RIS, and calculating the position of the scatterer by adopting a non-line-of-sight component of the scatterer and the calculated position of the user; tracking the positions of a user and a scatterer by using an extended Kalman filter (EKF) algorithm; according to the position of a user and a scatterer in a next time slot predicted by an EKF algorithm, transmitted beam forming and reflected beam forming are jointly optimized, the average reachable rate of the user is maximized, and the method adopts a two-step positioning method based on compressed sensing by separating RIS from a non-line-of-sight path of the scatterer in a first time slot; and in a subsequent time slot, according to emission and reflection beam forming designed in a previous time slot, an EKF-based tracking algorithm is executed, so that centimeter-level positioning and tracking precision is realized in a user and scatterer positioning / tracking stage.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless communications, and in particular relates to a synaesthesia fusion method based on a RIS-assisted millimeter wave MIMO-OFDM system. Background Art

[0002] ISAC has been recognized as a key enabler for the upcoming 6G wireless communication system, which can provide higher data rates, ultra-reliable and low-latency communications, and ubiquitous connectivity. Positioning and environmental awareness, as an integral aspect of perception, can support a variety of emerging usage scenarios, such as virtual reality, automated factories, and low-altitude economy. Millimeter wave communications and MIMO technology can provide high temporal and spatial resolution, paving the way for high-precision positioning and environmental perception.

[0003] The most common measurement information for positioning is arrival time (TOA), departure angle (AOD), arrival angle (AOA) and received signal strength (RSS). In the past, non-line-of-sight components were usually regarded as interference that needed to be eliminated. As the communication frequency evolves to the millimeter wave band, the sparsity of the millimeter wave channel causes the NLOS component to change its role. Recently, the use of NLOS components for positioning and environmental perception has been widely explored. For example, some technical solutions consider millimeter wave MIMO networks and propose a positioning and environmental perception algorithm based on soft information for location-assisted communication. In some technical solutions, the inventors use the joint sparsity of the millimeter wave channel in the angle and delay domains to propose an inexact block minimization-maximization positioning algorithm. However, due to the short wavelength of millimeter waves, a large amount of penetration loss and path loss lead to a decrease in positioning accuracy.

[0004] In order to overcome the above problems, RIS came into being. RIS is a plane composed of a large number of low-cost, low-power passive reflective elements, each of which can independently manipulate the amplitude and phase of the incident electromagnetic wave. Therefore, by creating additional virtual line-of-sight paths, RIS can reshape the wireless propagation environment in an energy-saving and economical way, achieve efficient reflection beamforming and promote millimeter-wave non-line-of-sight positioning. For MIMO-OFDM systems, some technical solutions derive the basic limitations of RIS-assisted positioning using a single base station, and some technical solutions study the RIS-assisted user tracking problem. Some inventors have established a probabilistic transfer model of user mobility and developed a message-passing-based method to track the user's location. In some technical solutions, a single-stage linear weighted least squares calculator is proposed to simultaneously determine the location of the user and the scatterer. Subsequently, a second-stage calculator is introduced to improve the environmental perception performance. However, the work in the prior art does not take advantage of the potential advantages of non-line-of-sight paths and does not consider the time correlation of user locations.

[0005] Based on the above technical problems existing in the prior art, the present invention proposes a synaesthesia fusion method based on a RIS-assisted millimeter wave MIMO-OFDM system. Summary of the invention

[0006] The purpose of the present invention is to provide a synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system in view of the deficiencies of the prior art.

[0007] The present invention adopts the following technical solution:

[0008] A synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system, comprising:

[0009] Step 1: Positioning the user and the scatterer. The user position is calculated using the non-line-of-sight component of the intelligent reflector RIS, and the scatterer position is calculated using the non-line-of-sight component of the scatterer and the calculated user position.

[0010] Step 2: Use the extended Kalman filter (EKF) algorithm to track the positions of the user and the scatterer;

[0011] Step 3: According to the user and scatterer positions of the next time slot predicted by the EKF algorithm, the transmit beamforming and the reflective beamforming are jointly optimized to maximize the average arrival rate of the user.

[0012] Furthermore, in step 1, during the user and scatterer positioning phase, 2M pilot symbols are included, and in the first M symbols, the phase shift matrix at the RIS is set to Θ (t) , and set to -Θ in the remaining M symbols (t) , user and scatterer positioning is performed only in the first time slot (t=1), and two sets of received signals are obtained by separating the non-line-of-sight components of RIS and scatterer transmissions:

[0013]

[0014] Among them, ω (t) represents the transmit beamforming vector at the base station, and They represent the equivalent receiving noise of the two groups of signals respectively, and The received signals on the kth subcarrier representing the mth symbol and the m+Mth symbol respectively can be expressed as:

[0015]

[0016] in, and represents the received noise, which obeys the complex Gaussian distribution, and They represent the channels from the base station to the RIS, from the RIS to the user, and from the base station to the user through the lth scatterer:

[0017]

[0018] in,(·) H represents the conjugate transpose of the matrix, g R,B and τ R,B represents the complex channel gain and delay of the path from the base station to the RIS, and represents the complex channel gain and delay from RIS to user path, and represents the complex channel gain and delay from the base station to the user via the lth scatterer, and represents the AOD of the path from the base station to the RIS and from the base station to the lth scatterer, and represents the AOA of the path from RIS to the user and from the lth scatterer to the user, and represents the AOD (AOA) of RIS along the y and z directions, and are the steering vectors of the base station, user and RIS on the kth subcarrier respectively.

[0019] Furthermore, in step 1, using To calculate the user location, let Formula (1) is modified as follows:

[0020]

[0021] Stacking along m and k get:

[0022]

[0023] in, K represents the number of subcarriers, is a function of the user's location, represents the complex gain over all symbols, represents the equivalent receiving noise;

[0024] Discretize the user's area of ​​interest into a set of grid points Assume the user location is contained in In this paper, formula (9) is converted into a compressed sensing model:

[0025]

[0026] In formula (10), ΦU represents a sparse dictionary, X U is a sparse vector representing the user's location;

[0027] The user localization problem is transformed into a joint parameter dictionary learning and sparse signal recovery problem:

[0028]

[0029] Among them, ξ represents the acceptable reconstruction error, and the parameterized sparse Bayesian dictionary learning PSBDL algorithm is used to solve By calculation and To get the user's location

[0030] Furthermore, in step 1, and To calculate the scatterer position, convert formula (2) into:

[0031]

[0032] in,

[0033]

[0034]

[0035]

[0036]

[0037] The scatterer positioning problem is transformed into:

[0038]

[0039] Use the PSBDL algorithm to solve The scatterer position is determined by and to calculate.

[0040] Further, in step 2, in the tth time slot, where t≥2, the transmit and reflective beamforming designed in the t-1th time slot is performed, and the EKF is used to track the user and the scatterer, and the Represents the state vector, and uses represents the observation vector, where and denote the positions of the user and the lth scatterer respectively, Denotes the user speed, assuming that the user moves at a constant speed, and the scatterer position difference between any two adjacent time slots is modeled by independent Gaussian noise, the state transition model and the observation model are expressed as:

[0041]

[0042] Where F represents the state transfer matrix, h(·) is defined by formula (3), and represents the noise vector.

[0043] Furthermore, step 2 includes:

[0044] Step 2.1: User and scatterer state prediction, where the user and scatterer states include position and velocity:

[0045]

[0046] Step 2.2, linearize the measurement model:

[0047]

[0048] Step 2.3, the mean square error (MSE) matrix prediction of the user and scatterer states:

[0049] P (t|t-1) =FP (t-1) F T +Q……(18),

[0050] Step 2.4, Kalman gain calculation:

[0051]

[0052] Step 2.5, user and scatterer state (position and velocity) tracking:

[0053]

[0054] Step 2.6, the MSE matrix of the position and velocity of the user and scatterer states is updated:

[0055] P (t) =(IK (t) H (t) ) (t|t-1) ……(twenty two).

[0056] Further, in step 3, the received signal on the kth subcarrier in the tth time slot is expressed as:

[0057]

[0058] in, and yes As a function of , the average achievable rate maximization problem can be formulated as:

[0059]

[0060] Θ (t) =diag(θ (t) )……(26),

[0061]

[0062] Among them, P T Indicates the maximum transmit power at the base station.

[0063] Furthermore, in step 3, optimizing transmit beamforming includes:

[0064] definition The transmit beamforming optimization problem is expressed as:

[0065]

[0066] st(25),

[0067] Introduce two auxiliary variables a k and p k ,definition and can be equivalently transformed into:

[0068]

[0069] Due to the non-convex constraints (30) and (31), Still non-convex, use SDR to process formula (30) and define Yes (t) ≥0 and rank(Ω (t) )=1, at the same time, and in Using SCA to process formula (31), using the first-order Taylor series expansion, The feasible point obtained at the rth iteration It is approximated by its linear lower bound:

[0070]

[0071] Remove the rank 1 constraint, It can be approximated as the following convex optimization problem:

[0072]

[0073] stΩ (t) ≥0……(35),

[0074] tr(Ω (t) )≤P T ……(36),

[0075]

[0076]

[0077] The Gaussian randomization method was used to obtain Rank-one solution of .

[0078] Furthermore, in step 3, the reflection beamforming optimization includes:

[0079] definition as well as The reflection beamforming optimization problem can be formulated as:

[0080]

[0081] Introduce two auxiliary variables b k and q k ,definition and was reformulated as:

[0082]

[0083] SDR is used to handle non-convex constraints (27) and (42), and an auxiliary variable w is introduced, defined as as well as have as well as get in,

[0084]

[0085] Using SCA to process (43), we have Therefore, removing the rank 1 constraint, can be equivalently transformed into:

[0086]

[0087] Using Gaussian randomization method to obtain Rank 1 solution of .

[0088] The beneficial effects of the present invention are:

[0089] The synaesthesia fusion method based on the RIS-assisted millimeter wave MIMO-OFDM system of the present invention divides each time slot into three stages: user and scatterer positioning / tracking, joint beamforming optimization and data transmission;

[0090] In the user and scatterer positioning / tracking phase, a two-step positioning method based on compressed sensing is adopted by separating the non-line-of-sight paths of RIS and scatterers in the first time slot; in the subsequent time slots, an EKF-based tracking algorithm is executed according to the transmit and reflect beamforming designed in the previous time slot, so that the user and scatterer positioning / tracking phase achieves centimeter-level positioning and tracking accuracy;

[0091] In the joint beamforming optimization stage, the user and scatterer positions of the next time slot predicted in the EKF process are used to jointly optimize the transmit beamforming and reflective beamforming with the help of optimization tools such as AO, SDR and SCA to maximize the average achievable rate of the user and improve the communication performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 It is a flow chart of the synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system described in an embodiment of the present invention;

[0093] Figure 2 This is a scene diagram of the synaesthesia fusion method based on the RIS-assisted millimeter wave MIMO-OFDM system described in an embodiment of the present invention;

[0094] Figure 3 is a schematic diagram of the synaesthesia fusion protocol described in an embodiment of the present invention;

[0095] Figure 4 RMSE curve diagram of the user and scatterer positions of the two-step positioning method proposed in the user and scatterer positioning stage in the embodiment of the present invention under different numbers of subcarriers;

[0096] Figure 5 A tracking performance curve diagram of the EKF-based tracking method proposed in the user and scatterer tracking stage in an embodiment of the present invention under different conditions;

[0097] Figure 6 It is a convergence performance curve of the joint beamforming optimization scheme based on alternating optimization in an embodiment of the present invention and a curve diagram of the average achievable rate of users under different transmission power conditions. DETAILED DESCRIPTION

[0098] In order to more clearly understand the above-mentioned purposes, features and advantages of the present invention, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0099] Example

[0100] like Figure 1As shown, the synaesthesia fusion method based on the RIS-assisted millimeter-wave MIMO-OFDM system divides the entire cycle into several time slots. In each time slot, the AOA, AOD, time delay and complex channel gain of each non-line-of-sight path are set as constants. Each time slot is divided into three stages. The goal of the first stage is to locate / track the position of the user and the scatterer through pilot symbols. The goal of the second stage is to use the calculated position information to jointly optimize the transmission and reflection beamforming to maximize the user's achievable rate in the next time slot. The goal of the third stage is to transmit data symbols. It should be noted that user and scatterer positioning is only performed in the first time slot to obtain the initial position of the user and the scatterer, while user and scatterer tracking are performed in the remaining time slots.

[0101] Assume that there is a RIS-assisted MIMO-OFDM system, which consists of a base station, a user, a RIS, and L scatterers. The base station and the user have N B and N U antennas, arranged in a uniform linear array, RIS is equipped with N R =N Ry ×N Rz reflective elements, arranged in a uniform planar array, with the antenna and element spacing set to the carrier frequency wavelength λ c Half of the time is transmitted in the downlink, where the OFDM signal occupies K subcarriers and the subcarrier spacing is Δf. Assume that the line-of-sight path between the base station and the user is blocked by obstacles. The positions of the base station, RIS, the user in the tth time slot, and the lth scatterer are represented by p B 、p R , and like Figure 2 shown.

[0102] Exemplary preset related parameters in the embodiments of the present application:

[0103] Base station location p B =(0,0,8)m,RIS position p R =(20,40,12)m, the subcarrier frequency is 28GHz, the number of scatterers is L=3, and the number of antennas at the base station and the user is N respectively B =12,N U =6, the number of elements at RIS is N R =100, subcarrier spacing is Δf=120KHz, number of subcarriers K=16, number of symbols M=8, noise power is σ 2 = -80dBm, users and scatterers are randomly distributed in the region of interest, the speed of users in the x-direction and y-direction is 0.8m / s and 1m / s respectively, the length of each time slot is set to 0.1s, and there are 200 time slots in total;

[0104] In the tth time slot, the channels from the base station to the RIS, from the RIS to the user, and from the base station to the user through the lth scatterer on the kth subcarrier are given by:

[0105]

[0106] Among them, g R,B and τ R，B represents the complex channel gain and delay of the path from the base station to the RIS, and represents the complex channel gain and delay from RIS to user path, and represents the complex channel gain and delay from the base station to the user via the lth scatterer, and represents the AOD of the path from the base station to the RIS and from the base station to the lth scatterer, and represents the AOA of the path from RIS to the user and from the lth scatterer to the user, and represents the AOD (AOA) of RIS along the y and z directions, and are the steering vectors of the base station, user and RIS on the kth subcarrier respectively, and the phase shift matrix at RIS is represented as Θ (t) =diag(θ (t) ), θ (t) is the transmit beamforming vector, and the transmit beamforming vector at the base station is represented as ω (t) , the received signal on the kth subcarrier of the mth OFDM symbol is expressed as:

[0107]

[0108] Among them, x (t) Indicates the pilot signal, set x (t) = 1 for positioning and tracking, noise It obeys complex Gaussian distribution.

[0109] The synaesthesia fusion method based on the RIS-assisted millimeter wave MIMO-OFDM system includes:

[0110] Step 1, such as Figure 3As shown in the first stage of the initial time slot, the proposed two-step positioning method is used to locate the user and the scatterer. After the user receives the signal shown in formula (4), the signal component of the base station-RIS-user path and the signal component of the base station-scatterer-user path are separated by operation, as shown in formula (5) and formula (6). Then, in the first step, the compressed sensing technology is used to construct the problem The PSBDL algorithm is used to solve and obtain the calculation of the initial user location In the second step, with the help of the calculated Constructing the sparse recovery problem The PSBDL algorithm is used to solve the initial scatterer position calculation Among them, positioning performance such as Figure 4 As shown;

[0111] Step 2: In the first stage of the t-th time slot (t≥2), the EKF algorithm is used to track users and scatterers. The received signal in formula (4) is used to track users and scatterers according to the EKF steps (steps 2.1 to 2.6). The transmit beamforming and reflection beamforming in formula (4) are obtained by the joint beamforming optimization stage in the t-1th time slot. The tracking performance is as follows: Figure 5 As shown;

[0112] Step 3, in the second stage of the tth time slot (t≥1), the AO-based joint beamforming optimization method is used to design the transmit beamforming and reflection beamforming. The goal is to optimize the transmit beamforming and reflection beamforming according to the user and scatterer positions in the t+1th time slot predicted by the EKF step 2.1 to maximize the average achievable rate of the user and achieve real-time beam alignment; First, a joint optimization problem is formulated Next, AO technology is used to Decompose into transmit beamforming optimization problem and reflection beamforming optimization problem For each sub-problem and The SDR, SCA and Gaussian randomization methods are used to solve the problem respectively, where the communication performance is shown in Figure 6 ;

[0113] Wherein, step 1 comprises:

[0114] Step 1.1, the user and scatterer positioning stage includes 2M pilot symbols, where the phase shift matrix at the RIS is set to θ in the first M symbols and to -θ in the remaining M symbols. Since the user and scatterer positioning is performed only in the first time slot, the time slot index "t" is omitted for simplicity. The non-line-of-sight components transmitted through the RIS and scatterers are separated, and two new sets of received signals are obtained:

[0115]

[0116] Step 1.2, use the To calculate the user's location, by defining Rewrite formula (5) as:

[0117]

[0118] Stacking along m and k get:

[0119]

[0120] Inspired by the theory of compressed sensing, the user's area of ​​interest is discretized into a set of grid points Assume the user location is contained in Therefore, formula (8) is converted into the standard compressed sensing model:

[0121]

[0122] Among them, Φ U represents a sparse dictionary, X U is a sparse vector representing the user's location. However, the user's location may deviate from the predetermined grid point. Therefore, the user positioning problem can be transformed into a joint parameter dictionary learning and sparse signal recovery problem.

[0123]

[0124] Use the parameterized sparse Bayesian dictionary learning (PSBDL) algorithm to solve Then, by calculating and To obtain the user's location calculation

[0125] Step 1.3, in step 1.2, use and To calculate the scatterer position, formula (6) can be equivalently converted into

[0126]

[0127] in,

[0128]

[0129]

[0130] g U,B =[g U,B,1, …, g U,B,L ] T ,

[0131]

[0132] The scatterer localization problem can be transformed into:

[0133]

[0134] Use the PSBDL algorithm to solve The scatterer position can be determined by and to calculate.

[0135] Step 2 includes:

[0136] Step 2.1, state prediction:

[0137]

[0138] Step 2.2, linearization:

[0139]

[0140] Step 2.3, mean square error (MSE) matrix prediction:

[0141] P (t|t-1) =FP (t-1) F T +Q......(17),

[0142] Step 2.4, Kalman gain calculation:

[0143]

[0144] Step 2.5, status tracking:

[0145]

[0146] Step 2.6, MSE matrix update:

[0147] P (t) =(IK (t) H (t) ) (t|t-1) ……(twenty one).

[0148] Step 3 includes:

[0149] Step 3.1, the received signal on the kth subcarrier in the tth time slot can be expressed as:

[0150]

[0151] in, and yes Therefore, the average achievable rate maximization problem can be formulated as:

[0152]

[0153] Θ (t) =diag(θ (t) )……(23b),

[0154]

[0155] Among them, P T represents the maximum transmit power at the base station. Due to the intractable logarithmic objective function, the unit mode constraint in (23c), and ω (t) and θ (t) The coupling relationship between is non-convex and difficult to solve directly. Therefore, an efficient algorithm based on alternating optimization (AO) is used to Decompose it into two sub-problems and solve them alternately. For each sub-problem, optimization techniques such as semidefinite relaxation (SDR) and continuous convex approximation (SCA) are used.

[0156] Step 3.2, transmit beamforming optimization, for any given θ (t) , optimize transmit beamforming ω (t) ,definition The transmit beamforming optimization problem is expressed as:

[0157]

[0158]

[0159] Introduce two auxiliary variables a k and p k ,definition and therefore, can be equivalently transformed into:

[0160]

[0161] Due to the non-convex constraints (24a) and (24b), Still non-convex, to solve this problem, first use SDR to process (24a), define Yes (t) ≥0 and rank(Ω (t) )=1, at the same time, and in Next, using SCA to process equation (24b), using the first-order Taylor series expansion, The feasible point obtained at the rth iteration It is approximated by its linear lower bound:

[0162]

[0163] By removing the rank 1 constraint, can be approximated as the following convex optimization problem:

[0164]

[0165] stΩ (t) ≥0……(26a),

[0166] tr(Ω (t) )≤P T ...(26b),

[0167]

[0168]

[0169] is a standard semidefinite programming problem, which is effectively solved by the existing convex optimization solver. Finally, the Gaussian randomization method is used to obtain rank-one solution of ;

[0170] Step 3.3, reflection beamforming optimization, for any given ω (t) , optimize the reflection beamforming θ (t) ,definition as well as The reflection beamforming optimization problem can be formulated as:

[0171]

[0172] By introducing two auxiliary variables b k and q k ,definition and therefore, was reformulated as:

[0173]

[0174]

[0175] First, we use SDR to handle the non-convex constraints (23c) and (27c) by introducing an auxiliary variable w and defining as well as have as well as In addition, get in:

[0176]

[0177] Next, using SCA to process formula (27d), similar to step 3.2, we have Therefore, removing the rank 1 constraint, can be equivalently transformed into:

[0178]

[0179] is a convex semidefinite programming problem, which can also be solved by a convex optimization solver. Finally, the Gaussian randomization method is used to obtain Rank 1 solution of .

[0180] The present invention is not limited by the above embodiments. The above embodiments and descriptions are only for explaining the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, which are within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the appended claims.

Claims

1. A synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system, characterized in that: include: Step 1: Positioning the user and the scatterer. The user position is calculated using the non-line-of-sight component of the intelligent reflector RIS, and the scatterer position is calculated using the non-line-of-sight component of the scatterer and the calculated user position. Step 2: Use the extended Kalman filter (EKF) algorithm to track the positions of the user and the scatterer; Step 3: According to the user and scatterer positions of the next time slot predicted by the EKF algorithm, the transmit beamforming and the reflective beamforming are jointly optimized to maximize the average arrival rate of the user.

2. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 1 is characterized in that: In step 1, the user and scatterer are located, including 2M pilot symbols. In the first M symbols, the phase shift matrix at RIS is set to Θ (t) , and set to -Θ in the remaining M symbols (t) , user and scatterer positioning is performed only in the first time slot, where t = 1. By separating the non-line-of-sight components of RIS and scatterer transmissions, two sets of received signals are obtained: Among them, ω (t) represents the transmit beamforming vector at the base station, and They represent the equivalent receiving noise of the two groups of signals respectively, and Represent the received signals on the kth subcarrier of the mth symbol and the m+Mth symbol respectively: in, and represents the received noise, which obeys the complex Gaussian distribution, and They represent the channels from the base station to the RIS, from the RIS to the user, and from the base station to the user through the lth scatterer: in,(·) H represents the conjugate transpose of the matrix, g R,B and τ R,B represents the complex channel gain and delay from the base station to the RIS path, and represents the complex channel gain and delay from RIS to user path, and represents the complex channel gain and delay from the base station to the user via the τth scatterer, θ R,B and represents the AOD of the path from the base station to the RIS and from the base station to the lth scatterer, and represents the AOA of the path from RIS to the user and from the lth scatterer to the user, and represents the AOD (AOA) of RIS along the y and z directions, and are the steering vectors of the base station, user and RIS on the kth subcarrier respectively.

3. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 2 is characterized in that: In step 1, user positioning is performed using To calculate, let Formula (1) is modified as follows: Stacking along m and k get: in, K represents the number of subcarriers, is a function of the user's location, represents the complex gain over all symbols, represents the equivalent receiving noise; Discretize the user's area of ​​interest into a set of grid points Assume the user location is contained in In this paper, formula (9) is converted into a compressed sensing model: In formula (10), Φ U represents a sparse dictionary, X U is a sparse vector representing the user's location; The user localization problem is transformed into a joint parameter dictionary learning and sparse signal recovery problem: Among them, ξ represents the acceptable reconstruction error, and the parameterized sparse Bayesian dictionary learning PSBDL algorithm is used to solve By calculation and To get the user's location 4. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 3 is characterized in that: In step 1, the scatterer is positioned using and To calculate, convert formula (2) into: in, The scatterer location is converted into: Use the PSBDL algorithm to solve The scatterer position is determined by and to calculate.

5. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 2 is characterized in that: In step 2, in the tth time slot, where t ≥ 2, the transmit and reflect beamforming designed in the t-1th time slot is performed, and the EKF is used to track the user and scatterer, and the Represents the state vector, and uses represents the observation vector, where and denote the positions of the user and the lth scatterer respectively, Denotes the user speed, assuming that the user moves at a constant speed, and the scatterer position difference between any two adjacent time slots is modeled by independent Gaussian noise, the state transition model and the observation model are expressed as: Where F represents the state transfer matrix, h(·) is defined by formula (3), and represents the noise vector.

6. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 1 is characterized in that: Step 2 includes: Step 2.1: User and scatterer state prediction, where the user and scatterer states include position and velocity: Step 2.2, linearize the measurement model: Step 2.3, prediction of the mean square error matrix of the user and scatterer states: P (t|t-1) =FP (t-1) FT+Q……(18), Step 2.4, Kalman gain calculation: Step 2.5, position and velocity tracking of user and scatterer states: Step 2.6, the MSE matrix of the user and scatterer states is updated: P (t) =(I-K (t) H (t) )P (t|t-1) ……(22)。 7. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 2 is characterized in that: In step 3, the received signal on the kth subcarrier in the tth time slot is expressed as: in, and yes As a function of , the average achievable rate maximization problem can be formulated as: Among them, P T Indicates the maximum transmit power at the base station.

8. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 1 is characterized in that: In step 3, transmit beamforming optimization includes: definition The transmit beamforming optimization problem is expressed as: Introduce two auxiliary variables a k and p k ,definition and can be equivalently transformed into: Due to the non-convex constraints (30) and (31), Still non-convex, use SDR to process formula (30) and define Yes (t) ≥0 and rank(Ω (t) )=1, at the same time, and in Using SCA to process formula (31), using the first-order Taylor series expansion, The feasible point obtained at the rth iteration It is approximated by its linear lower bound: Remove the rank 1 constraint, It can be approximated as the following convex optimization problem: The Gaussian randomization method was used to obtain Rank-one solution of .

9. The synaesthesia fusion method based on RIS-assisted millimeter wave MIMO-OFDM system according to claim 1, characterized in that: In step 3, the reflection beamforming optimization includes: definition as well as The reflection beamforming optimization problem can be formulated as: Introduce two auxiliary variables b k and q k ,definition and was reformulated as: SDR is used to handle non-convex constraints (27) and (42), and an auxiliary variable w is introduced, defined as as well as have as well as get in, Using SCA to process (43), we have Therefore, removing the rank 1 constraint, can be equivalently transformed into: Using Gaussian randomization method to obtain Rank 1 solution of .

Images