A sensor integration method for active intelligent reflector-assisted cloud access networks

Abstract

This invention discloses an integrated communication and sensing method for cloud access networks (CRAN) assisted by an active intelligent reflector. In the context of a CRAN, this method effectively reduces mutual interference between communication and sensing systems by leveraging the centralized signal processing advantages of a multi-antenna baseband processing unit (BBU). Simultaneously, it utilizes an active intelligent reflector (Active-IRS) to assist both communication and sensing. A joint design of a multi-antenna remote radio head (RRH) and beamforming at the active intelligent reflector ensures that the reflected signal from the active intelligent reflector is aligned with the target direction. The amplifier integrated into the active intelligent reflector enhances the energy of the reflected signal to counteract the impact of multipath loss on signal strength. This method achieves the desired reflector beam while maintaining the user's signal-to-interference-plus-noise ratio (SNR).

Description

Technical Field

[0001] This invention relates to the field of integrated communication and sensing technology, specifically to an integrated communication and sensing method for cloud access networks assisted by active intelligent reflective surfaces. Background Technology

[0002] In the post-5G and post-6G era of future communications, a series of emerging application scenarios will arise, including autonomous driving, industrial automation, and virtual reality. These scenarios place demands on systems beyond just communication between devices; they also require devices to possess a certain level of environmental awareness. Simultaneously, with the advent of the Internet of Things (IoT) in the 5G era, the communication needs between massive numbers of devices further increase the demand for spectrum resources in wireless communication networks. Considering the limited spectrum resources of wireless communication networks and the relatively close frequency bands used by communication and sensing systems, the need for spectrum sharing between these systems will further increase. This trend will inevitably lead to a series of problems such as spectrum congestion between the two systems and a decrease in communication system throughput. In summary, integrated communication and sensing technology can further integrate communication and sensing systems, ensuring communication quality while enabling the system to possess a certain level of sensing capability, thus alleviating the problem of spectrum resource scarcity. Therefore, integrated communication and sensing technology is considered one of the key candidate technologies for the post-5G and even 6G era. Summary of the Invention

[0003] The purpose of this invention is to achieve the simultaneous use of communication signals for sensing through integrated beamforming methods, thus realizing the goal of integrated communication and sensing. Simultaneously, it utilizes intelligent reflector technology to sense obscured targets, expanding the sensing range. In the context of cloud access networks, this invention, with the assistance of active intelligent reflectors, enables the simultaneous use of communication signals to sense targets in a specific direction, achieving the fusion of communication and sensing systems and effectively reducing interference and spectrum pollution between the two systems.

[0004] The technical solution of the present invention is as follows:

[0005] A sensing-communication integrated method for cloud access networks with active intelligent reflector assistance is proposed. The BBU pool generates communication and radar signals for L RRHs (Remotely Receiving Headquarters), which are transmitted to each RRH via a wired fronthaul link. Due to obstructions between the RRHs, the detection target, and the communication users, the active intelligent reflector assistance enables simultaneous target detection and the establishment of cascaded links between the RRHs and each communication user. The method involves joint optimization of the covariance and beamforming matrices of the beamforming matrices of all RRH transmitted signals, the reflection matrix of the active reflector, and the fronthaul quantization noise covariance matrix. Specifically, the method includes the following steps:

[0006] 1.1) In an integrated communication and sensing system assisted by an active intelligent reflector for cloud access networks, the system is based on a cloud access network architecture. It utilizes a BBU pool to centrally process communication and sensing data, distributing the centrally processed signal to each RRH (Resonant Heading Unit) via a wired fronthaul link. The active intelligent reflector can establish a virtual line-of-sight link to sense obscured targets, while simultaneously enhancing the amplitude of the reflected signal, effectively combating path loss, and increasing the signal energy illuminating the direction of the sensed target. The BBU pool simultaneously serves K single-antenna users and senses I targets through L multi-antenna RRHs. Each RRH has N... R One transmitting antenna. An active smart reflector has M reflective elements, which simultaneously assist in communication and sensing.

[0007] 1.2) The BBU pool first generates a signal. Where c is the data symbol for each user, W c The communication beamforming matrix is ​​defined, where x0 represents the sensing signal. This corresponds to the signal of the l-th RRH. Due to the limited capacity of the fronthaul link, it is necessary to adjust the signal... The signal is quantized and compressed before being transmitted to each RRH. After quantization and compression, the signal received by each RRH is:

[0008] x = W c c + x0 + q

[0009] Where c~CN(0,I) K x0 ~ CN(0,R0), where R0 is the covariance matrix of the sensed signal; q ~ CN(0,Ω), where Ω is the covariance matrix of the quantization noise. Due to point-to-point compression, this covariance matrix is ​​a block diagonal matrix. For signal x, its covariance matrix can be expressed as:

[0010]

[0011] 1.3) Each RRH is processed through a matrix Select the appropriate signal from the broadcast signal x for forwarding. Where matrix E... l In the (l-1)N R +1 to the lNth R The rows are identity matrices, with all remaining elements being 0. The signal transmitted by each RRH can be represented as:

[0012]

[0013] q l ~CN(0,Ω) l,l ) independent of The quantization noise, based on the property of the block diagonal of the quantization noise covariance matrix caused by point-to-point compression, can be obtained as Ω = diag({Ωl,l Each RRH receives a power constraint.

[0014] 1.4) The reflected signal form of an active reflecting surface is:

[0015] y IRS =ΨGx+Ψn+n s ,

[0016] Where Ψ is the reflection matrix, and based on the characteristic that an active reflector can amplify signals, the amplitude of each term in Ψ can be greater than 1. G is the channel between RRH and the reflector. The second term is the dynamic noise. The third factor is static noise. Dynamic noise is much greater than static noise, therefore static noise can be ignored. Active smart reflectors are subject to power limitations. The initialization of the reflector matrix Ψ needs to satisfy the power condition.

[0017] 1.5) The RRH transmitted signal passes through the direct link h between the RRH and the user. k Reflection link g with active reflector k ΨG reaches user k, where g k This is the channel between the reflector and the user. k =h k +g k ΨG represents the cascaded channel. The user-received signal can be represented as:

[0018] y k =z k x+g k Ψn+n k ,

[0019] The Gaussian white noise received by the user.

[0020] Furthermore, for an integrated communication and sensing system assisted by an active intelligent reflector in the access network, under the constraints of RRH and active intelligent reflector power, user signal-to-interference-plus-noise ratio (SNR) constraints, and fronthaul link constraints, the covariance matrix R and beamforming matrix W of the transmitted signal are optimized. c The noise covariance matrix Ω and the reflector matrix Ψ are quantized to minimize the mean square error between the reflected beam and the ideal beam of the active intelligent reflector. The specific steps are as follows:

[0021] 2.1) The reflected beam of a smart reflector at any angle θ can be expressed as:

[0022] P(θ;Ψ;R)=a H (θ)(ΨGRG H Ψ H +ΨΨ H)a(θ),

[0023] in The direction vector d represents the angle θ. IRS λ represents the spacing between adjacent units of the active intelligent reflective surface, and λ is the wavelength.

[0024] 2.2) The mean square error of the reflected beam and the ideal beam d(θ) in 2.1) at all angles θ can be expressed as:

[0025]

[0026] Where α is the scaling factor and J is the number of sampling angles.

[0027] 2.3) Regarding user communication constraints, it is desirable for the signal-to-interference-plus-noise ratio (SINR) of each user to be greater than a specific threshold to ensure user communication quality. The SINR condition for each user k can be expressed as:

[0028]

[0029] Where Γ is the threshold, w i For W c The i-th column, w k For W c The k-th column represents the beamforming vector of the current k-th user.

[0030] 2.4) For the fronthaul link limitation, the capacity of the limited fronthaul link between each BBU and RRH is C. l , express With x l Based on the mutual information between them, the fronthaul link constraints for each link can be expressed as follows:

[0031]

[0032] 2.5) The optimization problem can be expressed as:

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039] R≥0,Ω≥0

[0040] Where ≥ indicates the positive semidefiniteness of the matrix.

[0041] Further, step 2.5) addresses the optimization problem. Due to the non-convexity of the objective function and constraints, and the coupling between the various optimization variables, the original problem needs to be decomposed into two sub-problems, solved through alternating optimization. The specific steps are as follows:

[0042] 3.1) Subproblem 1 is characterized by first fixing the reflector matrix Ψ, and using the covariance matrix R, Ω, and user beamforming matrix W. c To optimize the variables, we construct a convex subproblem and optimize the convex subproblem. The specific steps are as follows:

[0043] 3.1.1) The continuous convex approximation (SCA) method is used to transform the fronthaul link constraint into a constraint on R and Ω. l,l Convex constraints, inequalities exist

[0044]

[0045] Established, of which S l For any positive semi-definite matrix, the equality condition is only true if...

[0046]

[0047] Established at that time, S l The power constraints for each RRH need to be initialized according to the above formula, and updated at the beginning of each round of alternating optimization based on the optimization result R of the previous round. In summary, the fronthaul link constraints in 2.4) can be transformed into the following convex constraint form:

[0048]

[0049] 3.1.2) Transform the user SINR constraint into a convex constraint form with respect to R, and define... The user SINR limit in 2.3) can be converted to the following regarding R and R k Convex constraint form:

[0050]

[0051] 3.1.3) According to the definition It can also be converted into the following restriction form:

[0052]

[0053] Since R0 is the objective function and has no effect on the other constraints, omitting R0 yields the following relaxed constraint form:

[0054]

[0055] 3.1.4) Using the semidefinite relaxation (SDR) method to... Relaxing the rank-one constraint, we can obtain the optimization problem form of subproblem one as follows:

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062] R≥0,R k ≥0,Ω≥0

[0063] 3.1.5) Due to the optimization results obtained It doesn't have to be a rank-1 matrix; the optimal w can be obtained using the following formula. k and R k

[0064] R = R * ,

[0065] 3.2) Based on the optimization results R and w obtained in subproblem one k The characteristic of subproblem two is that the optimization variables R and Ω are fixed, and a convex form of subproblem two is constructed with the reflection surface matrix Ψ as the optimization variable. The specific steps are as follows:

[0066] 3.2.1) The user SINR condition in 2.3) can be equivalently expressed in subproblem two as follows:

[0067]

[0068] The left half of the formula is defined as the reachable rate R of user k. k .

[0069] 3.2.2) Signal y received by user k k Through a linear receiver u k Restore baseband signal minimize With the original baseband signal s k The mean square error between them can achieve the optimal forward transmission rate.

[0070]

[0071] in

[0072] 3.2.3)R k It can be rewritten as

[0073]

[0074] Where w k Let be a positive auxiliary variable. It has been proven that the optimal forward transmission rate satisfies the following conclusions.

[0075]

[0076]

[0077] Where ψ=diag(Ψ) H The constraints in 3.2.1) can be rewritten as follows:

[0078]

[0079] 3.2.4) Define the following variable processing

[0080]

[0081]

[0082]

[0083]

[0084]

[0085] in The above variables are the optimal w obtained from subproblem one before each subproblem two. k Update with R. The constraints can eventually be rewritten as about Convex form:

[0086]

[0087] in

[0088] 3.2.5) Rewrite about Power limitations of active smart reflectors:

[0089]

[0090] Where K = GRG H ,

[0091] 3.2.6) Rewrite about objective function

[0092]

[0093] in

[0094]

[0095] At the same time, auxiliary variables are used to constrain the optimization objective function, with restrictions.

[0096]

[0097] 3.2.7) Use semidefinite relaxation method directly on Optimization is performed to obtain the result ψ. A Taylor expansion is then used to obtain the lower bound of the objective function with respect to ψ. The objective function is then bounded by ψ. t Expand at the point, calculate the variables,

[0098]

[0099] get Where ψ t The initial value is obtained by satisfying the power condition of the active reflector and by the optimal ψ obtained from the previous round of Gaussian randomization during the alternating optimization process.

[0100] 3.2.8) Optimize the following sub-problems.

[0101]

[0102]

[0103]

[0104]

[0105] ψ H Q(θ)ψ-αd(θ)≤t j ,t j ≥0

[0106]

[0107]

[0108]

[0109]

[0110]

[0111]

[0112] 3.2.9) Based on the optimization obtained in step 3.2.8) Perform Gaussian randomization a sufficient number of times to obtain the best-performing variable ψ, and determine whether the optimization has converged. If it has not converged, update the variable S according to 3.1.1). l Optimize subproblem one in 3.1.4) and update variable w according to 3.1.5). k and R k Then, based on 3.2.4) and 3.2.7), update the relevant variables and optimize the subproblem in 3.2.8) until convergence.

[0113] The beneficial effects of this invention are: by utilizing the centralized information processing of the cloud access network, mutual interference between communication and sensing systems is effectively reduced; active intelligent reflectors are used to expand the sensing range, enabling the sensing of obscured targets; simultaneously, the amplification effect of the active intelligent reflectors effectively counteracts path loss, enhances signal energy in the direction of the sensing target, and improves sensing performance. Integrated beamforming achieves communication and sensing integration, enabling the sensing of targets in specific directions while ensuring the signal-to-noise ratio for communication users. This integrated communication and sensing design reduces the occupancy of limited frequency bands by two previously independent systems, providing an effective solution to alleviate the large frequency band demands of future 5G communication. Attached Figure Description

[0114] Figure 1 This invention integrates beamforming schemes, and presents a comparison diagram of the proposed schemes with different numbers of active intelligent reflective surface units and a sensing-only scheme;

[0115] Figure 2 A schematic diagram of the system of this invention. Detailed Implementation

[0116] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0117] A sensing-communication integrated method for cloud access networks with active intelligent reflector assistance is proposed. The BBU pool generates communication and radar signals for L RRHs (Remotely Receiving Headquarters), which are transmitted to each RRH via a wired fronthaul link. Due to obstructions between the RRHs, the detection target, and the communication users, the active intelligent reflector assistance enables simultaneous target detection and the establishment of cascaded links between the RRHs and each communication user. The method involves joint optimization of the covariance and beamforming matrices of the beamforming matrices of all RRH transmitted signals, the reflection matrix of the active reflector, and the fronthaul quantization noise covariance matrix. Specifically, the method includes the following steps:

[0118] 1.1) In an active intelligent reflector-assisted communication and sensing integrated system for cloud access networks, the BBU pool simultaneously serves K single-antenna users and senses a certain number of targets through L multi-antenna RRHs. Each RRH has N R One transmitting antenna. An active smart reflector has M reflective elements, which simultaneously assist in communication and sensing.

[0119] 1.2) The BBU pool first generates a signal. Where c is the data symbol for each user, W c The communication beamforming matrix is ​​defined, where x0 represents the sensing signal. This corresponds to the signal of the l-th RRH. Due to the limited capacity of the fronthaul link, it is necessary to adjust the signal... The signal is quantized and compressed before being transmitted to each RRH. After quantization and compression, the signal received by each RRH is:

[0120] x = W c c + x0 + q

[0121] Where c~CN(0,I) K x0 ~ CN(0,R0), where R0 is the covariance matrix of the sensed signal; q ~ CN(0,Ω), where Ω is the covariance matrix of the quantization noise. Due to point-to-point compression, this covariance matrix is ​​a block diagonal matrix. For signal x, its covariance matrix can be expressed as:

[0122]

[0123] 1.3) Each RRH is processed through a matrix Select the appropriate signal from the broadcast signal x for forwarding. Where matrix E... l In the (l-1)N R +1 to the lNth R The rows are identity matrices, with all remaining elements being 0. The signal transmitted by each RRH can be represented as:

[0124]

[0125] q l ~CN(0,Ω) l,l ) independent of The quantization noise, based on the property of the block diagonal of the quantization noise covariance matrix caused by point-to-point compression, can be obtained as Ω = diag({Ω l,l Each RRH receives a power constraint.

[0126] 1.4) The reflected signal form of an active reflecting surface is:

[0127] y IRS=ΨGx+Ψn+n s ,

[0128] Where Ψ is the reflection matrix, and based on the characteristic that an active reflector can amplify signals, the amplitude of each term in Ψ can be greater than 1. G is the channel between RRH and the reflector. The second term is the dynamic noise. The third factor is static noise. Dynamic noise is much greater than static noise, therefore static noise can be ignored. Active smart reflectors are subject to power limitations. The initialization of the reflector matrix Ψ needs to satisfy the power condition.

[0129] 1.5) The RRH transmitted signal passes through the direct link h between the RRH and the user. k Reflection link g with active reflector k ΨG reaches user k, where g k This is the channel between the reflector and the user. k =h k +g k ΨG represents the cascaded channel. The user-received signal can be represented as:

[0130] y k =z k x+g k Ψn+n k ,

[0131] The Gaussian white noise received by the user.

[0132] Step 1.1) proposes an integrated communication and sensing system for access networks with active intelligent reflector assistance. Under the constraints of RRH and active intelligent reflector power, user signal-to-interference-plus-noise ratio (SNR) limitations, and fronthaul link limitations, it optimizes the covariance matrix R and beamforming matrix W of the transmitted signal. c The noise covariance matrix Ω and the reflector matrix Ψ are quantized to minimize the mean square error between the reflected beam and the ideal beam of the active intelligent reflector. The specific steps are as follows:

[0133] 2.1) The reflected beam of a smart reflector at any angle θ can be expressed as:

[0134] P(θ;Ψ;R)=a H (θ)(ΨGRG H Ψ H +ΨΨ H )a(θ),

[0135] in The direction vector d represents the angle θ. IRS λ represents the spacing between adjacent units of the active intelligent reflective surface, and λ is the wavelength.

[0136] 2.2) The mean square error of the reflected beam and the ideal beam d(θ) in 2.1) at all angles θ can be expressed as:

[0137]

[0138] Where α is the scaling factor and J is the number of sampling angles.

[0139] 2.3) Regarding user communication constraints, it is desirable for the signal-to-interference-plus-noise ratio (SINR) of each user to be greater than a specific threshold to ensure user communication quality. The SINR condition for each user k can be expressed as:

[0140]

[0141] Where Γ is the threshold, w i For W c The i-th column, w k For W c The k-th column represents the beamforming vector of the current k-th user.

[0142] 2.4) For the fronthaul link limitation, the capacity of the limited fronthaul link between each BBU and RRH is C. l , express With x l Based on the mutual information between them, the fronthaul link constraints for each link can be expressed as follows:

[0143]

[0144] 2.5) The optimization problem can be expressed as:

[0145]

[0146]

[0147]

[0148]

[0149]

[0150]

[0151] R≥0,Ω≥0

[0152] Where ≥ indicates the positive semidefiniteness of the matrix.

[0153] The optimization problem proposed in step 2.5) requires decomposition into two sub-problems due to the non-convexity of the objective function and constraints, as well as the coupling between the optimization variables. These sub-problems are solved through alternating optimization. The specific steps are as follows:

[0154] 3.1) Subproblem 1 is characterized by first fixing the reflector matrix Ψ, and using the covariance matrix R, Ω, and user beamforming matrix W. c To optimize the variables, we construct a convex subproblem and optimize the convex subproblem. The specific steps are as follows:

[0155] 3.1.1) The continuous convex approximation (SCA) method is used to transform the fronthaul link constraint into a constraint on R and Ω. l,l Convex constraints, inequalities exist

[0156]

[0157] Established, of which S l For any positive semi-definite matrix, the equality condition is only true if...

[0158]

[0159] Established at that time, S l The power constraints for each RRH need to be initialized according to the above formula, and updated at the beginning of each round of alternating optimization based on the optimization result R of the previous round. In summary, the fronthaul link constraints in 2.4) can be transformed into the following convex constraint form:

[0160]

[0161] 3.1.2) Transform the user SINR constraint into a convex constraint form with respect to R, and define... The user SINR limit in 2.3) can be converted to the following regarding R and R k Convex constraint form:

[0162]

[0163] 3.1.3) According to the definition It can also be converted into the following restriction form:

[0164]

[0165] Since R0 is the objective function and has no effect on the other constraints, omitting R0 yields the following relaxed constraint form:

[0166]

[0167] 3.1.4) Using the semidefinite relaxation (SDR) method to... Relaxing the rank-one constraint, we can obtain the optimization problem form of subproblem one as follows:

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174] R≥0,R k ≥0,Ω≥0

[0175] 3.1.5) Due to the optimization results obtained It doesn't have to be a rank-1 matrix; the optimal w can be obtained using the following formula. k and R k

[0176] R = R * ,

[0177] 3.2) Based on the optimization results R and w obtained in subproblem one k The characteristic of subproblem two is that the optimization variables R and Ω are fixed, and a convex form of subproblem two is constructed with the reflection surface matrix Ψ as the optimization variable. The specific steps are as follows:

[0178] 3.2.1) The user SINR condition in 2.3) can be equivalently expressed in subproblem two as follows:

[0179]

[0180] The left half of the formula is defined as the reachable rate R of user k. k .

[0181] 3.2.2) Signal y received by user k k Through a linear receiver u k Restore baseband signal minimize With the original baseband signal s k The mean square error between them can achieve the optimal forward transmission rate.

[0182]

[0183] in

[0184] 3.2.3)R k It can be rewritten as

[0185]

[0186] Where w k Let be a positive auxiliary variable. It has been proven that the optimal forward transmission rate satisfies the following conclusions.

[0187]

[0188]

[0189] Where ψ=diag(Ψ) H The constraints in 3.2.1) can be rewritten as follows:

[0190]

[0191] 3.2.4) Define the following variable processing

[0192]

[0193]

[0194]

[0195]

[0196]

[0197] in The above variables are the optimal w obtained from subproblem one before each subproblem two. k Update with R. The constraints can eventually be rewritten as about Convex form:

[0198]

[0199] in

[0200] 3.2.5) Rewrite about Power limitations of active smart reflectors:

[0201]

[0202] Where K = GRG H ,

[0203] 3.2.6) Rewrite about objective function

[0204]

[0205] in

[0206]

[0207] At the same time, auxiliary variables are used to constrain the optimization objective function, with restrictions.

[0208]

[0209] 3.2.7) Use semidefinite relaxation method directly on Optimization is performed to obtain the result ψ. A Taylor expansion is then used to obtain the lower bound of the objective function with respect to ψ. The objective function is then bounded by ψ. t Expand at the point, calculate the variables,

[0210]

[0211] get Where ψ t The initial value is obtained by satisfying the power condition of the active reflector and by the optimal ψ obtained from the previous round of Gaussian randomization during the alternating optimization process.

[0212] 3.2.8) Optimize the following sub-problems.

[0213]

[0214]

[0215]

[0216]

[0217] ψ H Q(θ)ψ-αd(θ)≤t j ,t j ≥0

[0218]

[0219]

[0220]

[0221]

[0222]

[0223]

[0224] 3.2.9) Based on the optimization obtained in step 3.2.8) Perform Gaussian randomization a sufficient number of times to obtain the best-performing variable ψ, and determine whether the optimization has converged. If it has not converged, update the variable S according to 3.1.1). l Optimize subproblem one in 3.1.4) and update variable w according to 3.1.5). k and R k Then, based on 3.2.4) and 3.2.7), update the relevant variables and optimize the subproblem in 3.2.8) until convergence.

Claims

1. A sensing integration method for cloud access networks with active intelligent reflective surface assistance, characterized in that: The BBU pool generates communication and radar signals for L RRHs, which are transmitted to each RRH via wired fronthaul links. Due to obstructions between the RRHs, the detection targets, and the communication users, the detection of targets is achieved with the assistance of active intelligent reflectors, while simultaneously establishing cascaded links between the RRHs and each communication user. The covariance matrix and beamforming matrix of the beamforming matrix of all RRH transmitted signals, the reflection matrix of the active reflectors, and the fronthaul quantization noise covariance matrix are jointly optimized, specifically including the following steps: 1.1) In an active intelligent reflector-assisted communication and sensing integrated system for cloud access networks, the BBU pool simultaneously serves K single-antenna users and senses I targets through L multi-antenna RRHs. Each RRH has... One transmitting antenna, an active intelligent reflector with M reflective elements, simultaneously assists in communication and sensing; 1.2) The BBU pool first generates a signal. ,in Data symbols for each user, To shape the communication beam, To sense signals, This corresponds to the signal of the l-th RRH; due to the limited capacity of the fronthaul link, it is necessary to adjust the signal... After quantization and compression, the signal is transmitted to each RRH; after quantization and compression, the signal received by each RRH is: ； in ; , The covariance matrix of the sensed signal; , The covariance matrix for quantization noise, due to point-to-point compression, is a block diagonal matrix; for the signal... Its covariance matrix is ​​expressed as: ； 3) Each RRH is passed through a matrix Select the appropriate signal from the broadcast signal x for forwarding; where the matrix In the To the The row is a unit matrix, with all remaining elements being 0; the signal transmitted by each RRH is represented as: ； To be independent The quantization noise is obtained based on the property of the block diagonal of the quantization noise covariance matrix caused by the point-to-point compression mentioned above. Each RRH receives a power constraint. ; 1.4) The reflected signal form of an active reflector is: ； in The reflection matrix is ​​based on the signal amplification characteristics of an active reflector. The magnitude of each item is greater than 1; The first term represents the channel between the RRH and the reflector; the second term represents the dynamic noise. The third item is static noise, while dynamic noise is greater than static noise; therefore, static noise can be ignored. The active intelligent reflector is subject to power limitations. Reflection surface matrix The initialization needs to satisfy the power condition; 1.5) RRH transmits signals via a direct link between RRH and the user. Reflection link with active reflector Reaching user k, where This is the channel between the reflector and the user; This is a cascaded channel; the user-received signal is represented as: ； The user receives Gaussian white noise; 1.6) Under the constraints of RRH and active intelligent reflector-assisted communication and sensing integrated system for cloud access networks, including user signal-to-interference-plus-noise ratio (SINNR) limitations and fronthaul link limitations, the covariance matrix of the transmitted signal is optimized. Beamforming matrix for transmitting signals Quantization noise covariance matrix and reflector matrix This minimizes the mean square error between the reflected beam and the ideal beam of the active intelligent reflector.

2. The integrated sensing method for cloud access networks with active intelligent reflective surface assistance as described in claim 1, characterized in that, The specific steps in 1.6) are as follows: 2.1) For any angle The reflected beam of the intelligent reflector is represented as: ； in express The direction vector of the angle, The spacing between adjacent units of the active intelligent reflector. Wavelength; 2.2) At all angles The reflected beam and ideal beam in step 2.1) above The mean square error is expressed as: ； in Scaling factor The number of sampling angles; 2.3) Regarding user communication constraints, it is desirable for each user's signal-to-interference-plus-noise ratio (SINR) to be greater than a specific threshold to ensure user communication quality. The SINR condition for each user k is expressed as follows: ； in For the threshold, for The i-th column, for The k-th column represents the beamforming vector of the current k-th user; 2.4) Regarding the fronthaul link limitation, the capacity of the limited fronthaul link between each BBU and RRH is... , express and The mutual information between them yields the expression for the fronthaul link limitation of each link as follows: ； 2.5) The optimization problem is expressed as: ； in This indicates the positive semidefiniteness of the matrix.

3. The integrated sensing method for active intelligent reflector-assisted cloud access networks according to claim 2, characterized in that, in step 2.5), the optimization problem, due to the non-convexity of the objective function and constraints, and the coupling between the various optimization variables, needs to be decomposed into two sub-problems and solved by alternating optimization; the specific steps are as follows: 3.1) The characteristic of subproblem one is that the reflection surface matrix is ​​fixed first. With covariance matrix , and user beamforming matrix To optimize the variables, we construct a convex subproblem and optimize the convex subproblem. The specific steps are as follows: 3.1.1) The continuous convex approximation SCA method is used to transform the fronthaul link constraint into a condition related to... and With convex constraints, the following inequalities apply: ； Established, among which For any positive semi-definite matrix, the equality condition is only true if: ； Established at that time, The initialization needs to satisfy the power constraints of each RRH according to the above formula, and at the beginning of each round of alternating optimization, the optimization results of the previous round should be used. The update is performed; in summary, the fronthaul link constraints in step 2.4) are transformed into the following convex constraint form: ； 3.1.2) Convert the user SINR constraint into a condition related to... The convex constraint form is defined. The user SINR limit in step 2.3) is converted to the following regarding and Convex constraint form: ； 3.1.3) According to the definition , It also transforms into the following restriction form: ； because Since the objective function and other constraints are unaffected, they are omitted. The following relaxed constraint form is obtained: ； 3.1.4) Using the semidefinite relaxation SDR method to... With the rank-one constraint relaxed, the optimization problem form of subproblem one is obtained as follows: ； 3.1.5) Due to the optimization results obtained It is not necessarily a rank-1 matrix; the optimal value can be obtained using the following formula. and : ； 3.2) Obtain the optimization result based on subproblem one. and The characteristic of subproblem two is that the optimization variables are fixed. and With the reflector matrix To construct a convex subproblem 2 for optimizing the variables, the specific steps are as follows: 3.2.1) For the user SINR condition in step 2.3), the equivalent expression in subproblem two is: ； The left half of the formula is defined as the reachable rate of user k. ; 3.2.2) Signal received by user k Through a linear receiver Restore baseband signal ;minimize Compared with the original baseband signal The mean square error between them achieves the optimal forward transmission rate before the target forward transmission rate is reached. ； in ; 3.2.3) Rewritten as: ； in Let be a positive auxiliary variable; it has been proven that the optimal forward transmission rate satisfies the following conclusions: ； in The constraints in step 3.2.1) are rewritten as follows: ； 3.2.4) Define the following variable processing. ; ； in ; The above variables are optimally obtained from subproblem one before each subproblem two. and Updated; the constraints were ultimately rewritten as regarding Convex form: ； in ; 3.2.5) Rewrite about Power limitations of active smart reflectors: ； in ; 3.2.6) Rewrite about objective function ； in , ； At the same time, auxiliary variables are used to constrain the optimization objective function, with the following restrictions: ； 3.2.7) Use semidefinite relaxation method directly on Optimize to obtain results ; The objective function is obtained through Taylor expansion. The lower bound of the objective function; Expand at the point, calculate the variables, ； get ;in The initial value is obtained by satisfying the power condition of the active reflector, and the optimal value is obtained from the previous round of Gaussian randomization during the alternating optimization process. get; 3.2.8) Optimize the following sub-problems: ； 3.2.9) Based on the optimization obtained in step 3.2.8), Performing a sufficient number of Gaussian randomizations yields the best performance. Determine if the optimization has converged. If not, update the variables according to step 3.1.1). Optimize sub-problem one in step 3.1.4) and update the variables according to step 3.1.5). and Then, based on steps 3.2.4) and 3.2.7), the relevant variables are updated, and the subproblems in step 3.2.8) are optimized until convergence.

Images