An interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network

Abstract

The present invention discloses an interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network, including: constructing a resource optimization problem with the transmission power of macro / micro base stations as the objective; setting the initial precoding matrix of macro / micro base stations and the initial reflection coefficient of the intelligent reflecting surface; given the reflection coefficient of the current iteration, the macro base station obtains the Lagrange multiplier set by solving the dual problem of the problem and combines the channel state information obtained by the Lagrange multiplier set and the modulation symbols to be sent, and designs the precoding matrix according to the symbol-level precoding criterion; the macro base station decomposes the precoding matrix of the micro base station into a phase part and a power part; in this iteration, given the precoding matrix of macro / micro base stations, the macro base station calculates the received signal quality and cross-layer interference expression converted by macro / micro devices according to the symbol-level precoding criterion with respect to the reflection coefficient, and then obtains the reflection coefficient of the intelligent reflecting surface. This application can effectively utilize multi-user interference and cross-layer interference in heterogeneous networks.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication, and particularly to an interference processing method for an intelligent reflecting surface-assisted multi-cell heterogeneous network. Background Art

[0002] The global commercial deployment of the Fifth-Generation (5G) mobile communication network is in full swing. Especially in terms of connectivity, the Global System for Mobile Communications Alliance (GSMA) predicts that 5G will increase the number of connections from 200 million in 2020 to 1.8 billion in 2025. In addition, more than 1,000 specific industry applications will benefit from the advantages of 5G, such as high bandwidth, low latency, and strong connectivity. Although the spectral efficiency and system capacity of 5G communication systems are several times higher than those of existing communication systems, there are still some problems to be solved, such as the construction cost of infrastructure, the energy consumption of base stations, and the rapid growth of data-centric automated devices, which are very likely to exceed the carrying capacity of the current 5G network. With the emergence of these problems, the Beyond-5G (B5G) mobile communication system has gradually come into the view of industrial and academic researchers, and at the same time, it has brought some new requirements and challenges, such as new performance indicators like higher throughput and energy efficiency, ultra-low latency and energy consumption. However, the rapid increase in wireless devices and the expansion of the access scale, combined with the differentiated data transmission requirements of various devices, pose great challenges to further improving the system performance of wireless communication systems.

[0003] In addition, the 5G mobile communication network will also show the characteristics of densification and heterogeneity. The ultra-dense small cells in the network will be severely damaged by inter-cell interference (ICI), intra-cell, adjacent channel, self-interference, and inter-channel interference. Moreover, the concurrent operations of various types of wireless devices in some small cells lead to co-layer and cross-layer interference. This poses great challenges to future mobile communication systems. Therefore, it is very necessary to develop an advanced interference management and mitigation technology for the next-generation cellular technology to maintain high service quality and fairness among users in the cellular network.

[0004] Most of the existing studies on IRS and interference utilization technologies are for single-cell networks and cannot solve the cross-layer interference problems brought by multi-cell networks. In addition, when designing the reflection coefficient of IRS in the existing studies, only the overall reception quality of all users is considered, which may result in the situation that the QoS of individual users cannot be satisfied. Summary of the Invention

[0005] The objective of the present invention is to mitigate and utilize the complex cross-layer interference and co-layer interference in heterogeneous communication networks, and to provide an interference handling method for intelligent reflecting surface-assisted multi-cell heterogeneous networks.

[0006] To achieve the objective of the present invention, an interference handling method for intelligent reflecting surface-assisted multi-cell heterogeneous networks provided by the present invention is applied to a downlink heterogeneous communication network assisted by an intelligent reflecting surface. In this network, there are macro base stations, micro base stations, macro devices, and micro devices. The intelligent reflecting surface is deployed near the macro base station and controlled by the macro base station. The micro base stations and macro / micro devices are all within the macro cell covered by the macro base station. The method includes:

[0007] Step 1: According to the quality of service requirements of the served macro devices and micro devices, construct the transmit power minimization optimization problem P1 of the macro / micro base stations, and jointly optimize the precoding matrix of the macro / micro base stations and the reflection coefficient of the intelligent reflecting surface;

[0008] Step 2: Set the initial iteration index q = 0, and set the initial transmit precoding vector W of the macro base station (0) and transmit power the transmit precoding matrix V of the micro base station (0) and transmit power the reflection coefficient θ of the intelligent reflecting surface IRS (0) , q = q + 1;

[0009] Step 3: Given the reflection coefficient θ in the q-th iteration (q) , convert the transmit power minimization optimization problem P1 into a single-objective optimization problem P2; solve the dual problem P'3 of the single-objective optimization problem P2 to obtain the Lagrange multiplier, and calculate the optimal macro base station precoding matrix in the q-th iteration according to the optimality of the KKT conditions

[0010] Step 4: The optimal precoding vector v of the micro base station n includes a phase part and a power part, and its expression is where is a unit beam vector that satisfies By setting the gradient value of the Lagrangian function to zero, we get Solve the power part of the optimal precoding vector v n Construct a standard interference function, which has positivity and monotonicity, and iterate the interference function through an iterative power control method to obtain the optimal power part

[0011] Step 5: Given the calculated optimal macro base station precoding matrix and Calculate the reflection coefficient by the multi-gradient descent method based on the Riemannian manifold.​

[0012] The present invention provides an interference processing method based on an intelligent reflecting surface-assisted multi-cell heterogeneous network for a multi-cell downlink heterogeneous communication network. In this application, consider a multi-cell heterogeneous communication network including a multi-antenna macro base station, a multi-antenna micro base station, an intelligent reflecting surface (IRS) deployed on a building, which is controlled by the macro base station and assists the macro base station in communication, a single-antenna macro device, and a single-antenna micro device. The macro base station mainly serves macro devices, and the micro base station mainly serves micro devices. In this application, it is assumed that the micro base station and all micro devices are under the coverage of the macro base station, and all macro devices are not under the coverage of the micro base station. This assumption is reasonable because the transmission power of the micro base station is relatively small, and the coverage range is only 10 - 300 meters. Therefore, all micro devices will be subject to multi-user interference from the micro base station and cross-layer interference from the macro base station.

[0013] In the method proposed in this application, the precoding of the macro / micro base station and the reflection coefficient of the IRS need to be calculated through alternating iteration. In the initial stage of the iteration, an initial precoding matrix and reflection coefficient are set. In each iteration, given the reflection coefficient of the previous iteration, the precoding matrix of the macro / micro base station is calculated, and then given the precoding matrix of this iteration, the reflection coefficient is calculated until the algorithm converges or reaches the maximum number of iterations. The specific method flowchart is as Figure 1 .

[0014] In the method proposed in this application, the precoding matrix of the macro / micro base station and the reflection coefficient of the IRS will be designed as follows:

[0015] The macro / micro base station obtains the channel state information (CSI) of each communication link through channel estimation and feedback. The macro / micro base station can share each other's CSI information and system parameter information, and this information can be realized through optical fiber transmission.

[0016] The macro base station constructs a constructive interference region according to the symbol-level precoding criterion and the modulation symbol data to be transmitted, such as Figure 2 , point A corresponds to the standard M-PSK constellation point, and the corresponding sector is the constructive interference region. In Figure 2 , the vector represents the target signal, the vector represents the received signal, and the vector represents the interference signal. If the condition θ AB ≤ θ t, the superimposed signal of the target signal and the interference signal will fall within the constructive interference region. When the superimposed signal of the target signal and the interference signal of the device is within the constructive interference region, the interference signal will drive the power of the target signal) to exceed the detection threshold. In this way, the macro base station can transmit signals with lower power, thus playing a role in saving energy consumption. In this application, the precoding matrix W = [w1, w2,..., w K of the macro base station is designed using the symbol-level precoding criterion. The main purpose of the traditional linear precoding strategy is to enhance the power of the target signal and suppress interference. The symbol-level precoding criterion of the present invention designs the precoding matrix by combining CSI information and modulation symbol information, which can adjust the phase angle between the interference signal and the target signal to an acute angle, so that the received modulation symbols are further pushed into the correct detection region, converting harmful multi-user interference into beneficial interference, and making the received signals of each macro device fall within the constructive interference region, improving the performance of the system.

[0017] Since designing the precoding using the symbol-level precoding criterion requires obtaining CSI information and modulation symbol data, considering information security, it is assumed that the modulation symbol data of the macro base station is not shared with the micro base station. At this time, the micro base station cannot convert the cross-layer interference into beneficial interference and can only use the traditional precoding scheme to design its precoding matrix. The deployment and adjustment of the reflection coefficient of the IRS make it easier for the received signals of each macro device to fall within the constructive interference region, further reducing the transmission power of the macro base station. However, deploying the IRS will increase the additional interference to the micro devices. This application changes the wireless channel environment from the macro base station to the macro device and from the macro base station to the micro device by changing the reflection coefficient of the IRS, improving the signal reception quality of each macro device and alleviating the cross-layer interference received by the micro devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flowchart of an interference processing method for an intelligent reflecting surface-assisted multi-cell heterogeneous network provided by this application;

[0019] Figure 2 is a schematic diagram of the constructive interference region of M-PSK.

[0020] Figure 3 is a schematic diagram of an implementation environment provided by an embodiment of this application;

[0021] Figure 4 is a flowchart of precoding and reflection coefficient design based on power minimization provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The core of the present invention is to provide an interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network to effectively mitigate and utilize cross-layer interference and co-layer interference in heterogeneous communication scenarios. To make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings. Apparently, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0023] An interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network provided by the present invention is applied to a multi-cell downlink heterogeneous communication network and can mitigate and utilize complex cross-layer interference and co-layer interference in the heterogeneous communication network. The network includes a macro base station with N M antennas, a micro base station with N P antennas, an intelligent reflecting surface IRS with M reflecting units deployed near the macro base station, controlled by the macro base station and assisting the macro base station in communication, K single-antenna macro devices and N single-antenna micro devices. The macro base station mainly serves macro users (macro devices), and the micro base station mainly serves micro users (micro devices). The micro base station and micro users are both within the coverage area of the macro base station. The modulation symbol vector sent by the macro base station is s = [s1, s2,..., s K , where s k , k ∈ {1, 2,..., K} represents the modulation symbol that the macro base station will send to the k-th macro device. In some embodiments of the present invention, the method provided by the present invention is described through a two-cell heterogeneous mobile communication system, and the system structure is as Figure 3 . The steps of the method are as follows:

[0024] Step 1: According to the Quality of Service (QoS) requirements of serving macro devices and micro devices, taking the transmission powers of the macro base station and the micro base station as the objective functions, construct an optimization problem P1 for minimizing the transmission powers of the macro base station and the micro base station. The purpose is to minimize the transmission powers of the macro / micro base stations by optimizing the precoding matrices W = [w1, w2,..., w K and V = [v1, v2,..., v N of the IRS, and the reflection coefficients θ = [θ1, θ2,..., θ M . T In the expression, K, N, and M are the numbers of macro devices, micro devices, and IRS units respectively, [·] T is the transpose operation, w K represents the precoding vector of the macro base station for the K-th macro device, and v NDenote the precoding vector of the micro base station for the Nth micro-device as θ M Denote the reflection coefficient of the Mth reflecting element of the IRS. By using the weighted Chebyshev method, the transmit power minimization optimization problem P1 is transformed into a single-objective optimization problem P2.

[0025] Step 2: Set the initial iteration index q = 0. Set the initial macro base station transmit precoding vector W (0) and the transmit power The micro base station transmit precoding matrix V (0) and the transmit power The reflection coefficient θ of the IRS (0) .

[0026] Step 3: Let q = q + 1. Set the vector composed of the precoding vectors of the macro base station for each macro-device as The expression is [·] H is the conjugate transpose operation. Where define The real vector of is Where and are respectively The real and imaginary parts of. Given the reflection coefficient θ (q) at the qth iteration, substitute into the single-objective optimization problem P2 to obtain the optimization problem P3 corresponding to the two optimization variables of and V. Derive the Lagrangian function of the optimization problem P3 to obtain the dual problem P′3 of P3. P′3 is a standard convex problem and can be solved by standard convex optimization methods. In some embodiments of the present invention, the interior point method can be used to solve it. By solving the dual problem P′3, the Lagrange multipliers ψ = [ψ1, ψ2,..., ψ N T , ρ = [ρ1, ρ2] T , v = [v1, v2,..., v K T . According to the optimality of the KKT conditions, calculate the optimal at the qth iteration as follows

[0027]

[0028] where 0 < ξ1 < 1 represents the weight value for optimizing the macro base station transmit power, v k represents the Lagrange multiplier corresponding to the QoS constraint of the kth macro-device, ψ n represents the Lagrange multiplier corresponding to the QoS constraint of the nth micro-device, ρ1 and ρ2 respectively represent the Lagrange multipliers corresponding to the transmit power constraints of the macro base station and the micro base station, represents 2N​​M The identity matrix of dimension σ M，k represents the standard deviation of the Gaussian white noise of the k-th macro device, Γ M，k represents the minimum SINR (Signal to Interference plus Noise Ratio) requirement of the k-th macro device. represents the threshold angle of the modulation constellation points, obtained by where is the modulation order of phase shift keying (PSK). Φ n，k and are the matrix and vector composed of the real and imaginary parts of the channel gain vector and the modulation symbol information, and the specific expressions are as follows

[0029]

[0030]

[0031]

[0032]

[0033]

[0034] where Φ = diag(θ), G represents the complex channel gain vectors from the macro base station to the n-th micro device, from the intelligent reflecting surface IRS to the n-th micro device, from the macro base station to the k-th macro device, from the intelligent reflecting surface IRS to the k-th macro device, and from the macro base station to the intelligent reflecting surface IRS, respectively. represents a 1×2N M (k - 1)-dimensional row zero vector, represents the vector composed of the complex channel gain vectors from the macro base station and IRS to the n-th micro device, μ k represents the vector combined with the modulation symbol to be sent by the macro base station, μ R，k represents the real part of μk, μ I，k represents μ k the imaginary part of represents the vector composed of the complex channel gain vectors from the macro base station and IRS to the k-th macro device,

[0035] Once the optimal precoding matrix of the macro base station is obtained the next step is to determine the optimal precoding matrix of the micro base station.

[0036] Step 4: The optimal precoding vector v of the micro base station nIt can be composed of a phase part and a power part, and its expression is

[0037]

[0038] where is the optimal precoding vector of the femto - base station is the power part of denotes the optimal precoding vector of the femto - base station is the phase part of

[0039] is a unit beam vector, satisfying By setting the gradient value of the Lagrangian function to zero, we obtain The expression of

[0040]

[0041] where ξ2 represents the weight value for optimizing the femto - base station transmit power, g n denotes the channel state information CSI from the femto - base station to the nth micro - device, denotes the N P dimensional identity matrix

[0042] Next, solve for the power part of the optimal precoding vector v n Introduce the standard interference function, which has positivity and monotonicity, and its expression is as follows

[0043] p = I(p) (5)

[0044] where p = [p1, p2,..., p N is the power part corresponding to the femto - base station precoding vector, is the power part of the optimal precoding vector v of the femto - base station n I = [I1, I2,..., I N is the set of standard interference functions of the micro - devices with respect to p, denotes the standard interference function of the nth micro - device, and its expression is as follows

[0045]

[0046] where Γ P，n and respectively represent the minimum SINR requirement of the kth macro - device and the variance of Gaussian white noise, denotes the phase part of the optimal precoding vector of the femto - base station, p j is the precoding vector v of the femto - base station jThe power part. Since I(p) is positive and monotonic, It can be solved by an iterative power control method, and its iterative criterion is as follows:

[0047] p (l) = I(p (l-1) ), (7)

[0048] where p (l) represents the solution of the l-th iteration. When the initial power vector is a feasible solution, the iterative power control method converges to a unique fixed point p * , and the optimal precoding vector of the micro base station is obtained

[0049] Step 5: Given the calculated optimal macro and micro base station precoding matrices and By adjusting the reflection coefficient θ of the intelligent reflecting surface IRS, the received signal of the macro device is more likely to be located in the constructed interference area and the interference received by the micro device from the macro base station and IRS is mitigated, the received signal quality of the macro device and the micro device is improved, and thus the power consumption of the macro / micro base station is further saved.

[0050] To meet the interference utilization condition of the received signal of the macro device and mitigate the cross-layer interference received by the micro device from the macro base station and IRS, taking the SINR expression u of the macro device satisfying the interference utilization condition and the interference expression u m of the micro device as the optimization objectives, a multi-objective optimization problem about the reflection coefficient θ = [θ1, θ2,..., θ M T is constructed, where M represents the number of reflection units of the IRS. Let u = [u1, u2,..., u M , where represents the phase expression corresponding to θ m , and K + N objective functions are constructed. The first to K objective functions are functions f l (u) of the QoS constraint of the macro device under the interference utilization criterion, l = 1, 2,... K, and the (K + 1)-th to (K + N)-th objective functions are the interference functions f l (u) of the micro device received from the IRS, l = K + 1, K + 2,..., K + N, representing the objective function index. The l-th objective function expression is as follows:

[0051]

[0052] where

[0053]

[0054] In expression (8), a lDenote the modulation symbol \(s\) transmitted by the macro base station, the precoding matrix \(W\) of the macro base station, and the channel gain \(h\) from the IRS to the first macro device, and the vector \(b\) composed of the channel gain \(G\) from the macro base station to the IRS r，l , l Denote the modulation symbol \(s\) sent to the first macro device l , the precoding matrix \(W\) of the macro base station, the channel gain \(h\) from the macro base station to the first macro device l and the scalar \(\sigma\) composed of the modulation symbol \(s\) transmitted by the macro base station M，l and \(F\) M，l respectively represent the additive white Gaussian noise and the SINR requirement of the first macro device, \(l = 1,2,\cdots,K\); \(c\) l-K，k Denote the channel gain \(h\) from the IRS to the \(l - K\)th micro device RI，l-K , the channel gain \(G\) from the macro base station to the IRS, and the precoding vector \(w\) sent by the macro base station to the \(k\)th macro device k to form the vector \(d_l\) -K， \(k\) represents the scalar composed of the channel gain \(h\) from the macro base station to the \(l - K\)th micro device MI，l-K and the precoding vector \(w\) sent by the macro base station to the \(k\)th macro device k .

[0055] The absolute value term in the objective function (8) will hinder the further solution of the problem. In this application, an approximate expression of the absolute value term is obtained by using the log-sum-exp inequality. Then the vector \(u\) is mapped to an \(M\)-dimensional manifold space which is characterized as follows:

[0056]

[0057] where is an \(M\)-dimensional complex vector, and \(0_M\) is an \(M\)-dimensional zero matrix. \((\cdot)^{(H)}\) * represents the conjugate operation, \(T\) u \(S\) is the tangent space of the manifold space . Then, based on the manifold space, the Riemannian gradient of each objective function \(f\) l (u) is derived, and then a common Riemannian gradient is obtained by weighted summation. The common Riemannian gradient ensures that each objective function will decrease in each iteration. Before starting the iteration, set the initial point \(u_1\) at and calculate the initial search direction \(d_1\). The main steps in the \(i\)th iteration are as follows:

[0058] (1) Update the search point: First, update the current point on the manifold space , and the search point in the next iteration is:

[0059]

[0060] where Retr(·) is the retraction operation that maps a numerical value to the manifold space, is a step size, which can be obtained by using the Armijo backtracking line search method, and u i represents the search point at the i-th iteration, and d i represents the search direction at the i-th iteration.

[0061] (2) Common Riemannian gradient: Find an appropriate common gradient so that all objective functions share a common descent direction in the next iteration. Calculate the Riemannian gradient of each objective function as follows:

[0062]

[0063] where represents the Euclidean gradient of J l (u i+1 ), ⊙, (·) * respectively represent taking the real value, Hadamard product, and conjugate operation. Then, calculate the expression of the common Riemannian gradient as follows:

[0064]

[0065] where r l (·) represents the Gram - Schmidt orthogonalization operation of gradJ l (·), and α l，i+1 represents the weight factor of the l-th objective function at the (i + 1)-th iteration, which is calculated by the following formula

[0066]

[0067] where r j (u i+1 ) represents the Gram - Schmidt orthogonalization operation of the Riemannian gradient of the l-th objective function at the (i + 1)-th iteration.

[0068] (3) Update the search direction: The third step is to calculate the next search direction by the conjugate gradient method, and the update rule of the next search direction is

[0069]

[0070] where η i represents the Polak - Ribiere parameter, represents the output vector after the Riemannian transformation, and the specific operation is as follows:

[0071]

[0072] where Denote the conjugate of the search point in the (i + 1)-th iteration.

[0073] Step 6: Calculate the transmit power of the macro base station in the q-th iteration and the transmit power of the micro base station in the q-th iteration respectively through the precoding matrices of the macro / micro base stations obtained in the q-th iteration and the transmit power of the micro base station Then compare with the result of the previous iteration to determine whether the convergence condition is satisfied. The judgment criterion is as follows:

[0074]

[0075] where and respectively represent the transmit power of the macro base station and the transmit power of the micro base station in the (q - 1)-th iteration, ε represents the system convergence threshold, and “&&” represents the AND operation. If satisfied, end the iterative step; otherwise, repeat Steps 3, 4, and 5.

[0076] In summary, the foregoing embodiments of the present invention disclose an interference processing method for an intelligent reflecting surface-assisted multi-cell heterogeneous network. By using the symbol-level precoding technology to design the reflection coefficients of the macro / micro base stations and the IRS in the IRS-assisted heterogeneous network scenario, the transmit power of the macro / micro base stations is minimized under the communication requirement constraints of the macro / micro devices in the heterogeneous network.

[0077] An interference processing method for an intelligent reflecting surface-assisted multi-cell heterogeneous network provided by the foregoing embodiments of the present invention can alleviate multi-user interference and cross-layer interference in the multi-cell heterogeneous network. In this method, through the Riemannian manifold-based multi-gradient descent method, the received gain of each macro device and micro device can be improved in each iteration.

[0078] The method provided by the foregoing embodiments of the present invention can be applied in the heterogeneous communication network scenario. By optimizing the precoding matrices of the macro base station and the micro base station and the reflection coefficients of the reconfigurable intelligent surface, the multi-user interference and cross-layer interference in the heterogeneous network are effectively utilized, and the transmit power of the macro / micro base stations is reduced.

[0079] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent substitution methods and are all included in the protection scope of the present invention.

Claims

1. A method for interference processing based on intelligent reflecting surface assisted multi-cell heterogeneous network, characterized in that The method includes: Step 1: According to the quality of service requirements of the serving macro device and micro devices, construct the transmit power minimization optimization problem P1 of the macro / micro base stations, and jointly optimize the precoding matrices of the macro / micro base stations and the reflection coefficients of the intelligent reflecting surface; Step 2: Set the initial iteration index q = 0, and set the initial transmit precoding vector W of the macro base station (0) and transmit power The transmit precoding matrix V of the small base station (0) and transmit power The reflection coefficient θ of the intelligent reflecting surface IRS (0) ; Step 3: Given the reflection coefficient θ at the q-th iteration (q) , the transmit power minimization optimization problem P1 is converted into a single-objective optimization problem P2; solve the dual problem P3' of the single-objective optimization problem P2 to obtain the Lagrange multiplier, and calculate the optimal macro base station precoding matrix at the q-th iteration according to the optimality of the KKT conditions Step 4: Optimal precoding vector v of the pico base station n It includes a phase part and a power part, and its expression is where is a unit beam vector that satisfies By setting the gradient value of the Lagrangian function to zero, we obtain Solve for the optimal precoding vector v n The power part of Construct a standard interference function, which has positivity and monotonicity. Iterate the interference function through an iterative power control method to obtain the optimal power part Step 5: Given the calculated optimal macro base station precoding matrix and Calculate the reflection coefficient by the multi-gradient descent method based on the Riemannian manifold; Among them, designing the precoding matrix of the macro / micro base station includes: setting the vector composed of the precoding vectors of the macro base station for each macro device as The expression is [·] H is the conjugate transpose operation, defined as The real vector of is where and are respectively The real part and the imaginary part of; Given the reflection coefficient θ at the q-th iteration (q) Under the condition of, substitute into the single-objective optimization problem P2, obtain the optimization problem P3 corresponding to the two optimization variables of and V, derive the Lagrangian function of the optimization problem P3 to obtain the dual problem P3′ of the optimization problem P3, solve the dual problem P′3 of the single-objective optimization problem P2, and obtain the Lagrange multipliers ψ = [ψ1, ψ2,..., ψ N T , ρ = [ρ1, ρ2] T , v = [v1, v2,..., v k T Solve the dual problem P3′, and obtain the Lagrange multipliers ψ = [ψ1, ψ2,..., ψ N T , ρ = [ρ1, ρ2] T , v = [v1, v2,..., v K T , According to the optimality of the KKT conditions, obtain the optimal macro base station precoding matrix at the current iteration The specific expression is as follows:​​​​ where \(0 < \xi_1 < 1\) represents the weight value for optimizing the transmit power of the macro base station, \(v\) k represents the Lagrange multiplier of the QoS constraint corresponding to the \(k\)-th macro device, \(\psi\) n represents the Lagrange multiplier of the QoS constraint corresponding to the \(n\)-th micro device, \(\rho_1\) and \(\rho_2\) respectively represent the Lagrange multipliers of the transmit power constraints corresponding to the macro base station and the micro base station, represents \(2N\) M dimensional identity matrix, \(\sigma\) M,k represents the standard deviation of the Gaussian white noise of the \(k\)-th macro device, \(\Gamma\) M,k represents the minimum signal-to-interference-plus-noise ratio requirement of the \(k\)-th macro device, represents the threshold angle of the modulation constellation points, \(\Phi\) n,k and are the matrix and vector composed of the real and imaginary parts of the channel gain vector and the modulation symbol information; In step 4, the unit beam vector has the following expression: where ξ2 represents the weight value for optimizing the transmission power of the micro base station, and g n represents the CSI from the micro base station to the nth micro device, represents N P dimensional identity matrix; In step 4, for the optimal power part The expression of the introduced standard interference function is as follows: p = I(p) where p = [p1, p2, …, p N is the power part corresponding to the precoding vector of the femto base station, is the optimal precoding vector v n of the femto base station, and I = [I1, I2, …, I N is the set of standard interference functions of the micro device with respect to p, represents the standard interference function of the nth micro device, and its expression is as follows: Γ P,n and respectively represent the minimum QoS requirement of the k-th macro device and the variance of Gaussian white noise, represents the phase part of the optimal precoding vector of the micro base station, p ; and p j is the power part of the precoding vector v of the micro base station j ; Since I(p) is positive and monotonic, it is solved by an iterative power control method Its iterative criterion is as follows: p (l) = I(p (l-1) ), where p (l) and I(p (l-1) ) represents the solution of the l-th iteration. When the initial power vector is a feasible solution, the iterative power control method converges to a unique fixed point p * , and the optimal precoding vector of the pico base station is obtained 2. The interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network according to claim 1, wherein The macro base station and the micro base stations can share some information, and the information includes CSI and the QoS requirements of each macro / micro device. The information sharing is achieved by wireless or optical fiber means.

3. The interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network according to claim 1, characterized in that The method decouples the precoding matrix and the reflection coefficient through an alternating optimization method, including: given the reflection coefficient, designing the precoding matrix of the macro / micro base stations; given the precoding matrix, designing the reflection coefficient of the IRS.

4. The interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network according to claim 1, wherein In the q-th iteration, given the optimal precoding matrix of the macro / micro base stations in the current iteration, adjust the reflection coefficient θ of the IRS, so that the received signal of the macro device is more likely to be located in the constructed interference region and mitigate the interference of the micro device from the macro base station and the IRS, thereby further saving the power consumption of the macro / micro base stations.

5. A method for interference processing in an intelligent reflecting surface-assisted multi-cell heterogeneous network according to any one of claims 1-4, characterized in that The SINR expression \(u\) that satisfies the interference utilization condition with the macro device and the interference expression \(u\) suffered by the micro device m As the optimization objective, construct a multi-objective optimization problem regarding the reflection coefficient \(\theta = [\theta_1, \theta_2, \ldots, \theta M T , where \(M\) represents the number of reflection units of the IRS. Let \(u = [u_1, u_2, \ldots, u M and Solve \(u\) based on the Riemannian manifold multi-gradient descent method.​ 6. The interference processing method based on intelligent reflecting surface assisted multi-cellular heterogeneous network according to claim 5, wherein, Map the vector u to an M - dimensional manifold space S M , and then derive the Riemannian gradient of each objective function based on the manifold space. A common Riemannian gradient is obtained through weighted summation, and it is ensured that each objective function decreases in each iteration.

7. The interference processing method based on intelligent reflecting surface assisted multi-cell heterogeneous network according to claim 6, wherein The iterative steps are as follows: Set the initial point and calculate the initial search direction; Update the search point on the manifold space; Find a common Riemannian gradient so that all objective functions share a common descent direction in the next iteration, the common Riemannian gradient The calculation formula of where r l (·) represents gradJ l (·) is the Gram-Schmidt orthogonalization operation, and α l,i+1 represents the weight factor of the l-th objective function at the (i + 1)-th iteration; Calculate the next search direction by the conjugate gradient method.

Images