NOMA (Non-Orthogonal Multiple Access)-assisted cellular-elimination and inductance-elimination integrated system beamforming method
By introducing NOMA user pairing and beamforming design in the ISAC system, the problem of performance loss in burst terminal access scenarios is solved, the system's access capability and service performance is improved, and multiple access needs and perception services are met. Improvement of quality.
Patent Information
- Application Number
- CN202411020768.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-22
AI Technical Summary
The existing ISAC system has suffered serious performance losses in burst terminal access scenarios, and the NOMA-assisted CFMM-ISAC system design has not been studied, and user clustering or pairing strategies have not been considered, which affects system performance.
A NOMA-assisted decellular synesthesized integrated system beamforming method is proposed. By combining NOMA technology with CFMM-ISAC technology, adaptive NOMA user pairing and beamforming design are adopted to introduce user pairing factors to maximize system communication Speed and ensure perceived service quality.
It significantly improves the user access capability and synesthesia integrated service performance of the CFMM-ISAC system, meets the needs of burst terminal access scenarios, and improves the number of connections and bandwidth efficiency of the system.
Smart Images

Figure CN119010972A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of signal processing in wireless communications, and in particular relates to a NOMA-assisted de-cellularized synaesthesia integrated system beamforming method. Background Art
[0002] Emerging services such as artificial intelligence services, immersive interactive services, and digital twin services are constantly emerging and are penetrating deeply into vertical application fields such as intelligent transportation, smart medical care, and intelligent manufacturing, which puts forward higher demands on the information processing capabilities of 6G networks. Integrated Sensing and Communication (ISAC) realizes the unified design of communication and perception functions through joint design of air interface and protocol, reuse of time-frequency and space resources, and hardware sharing, so that wireless networks can have high-precision and refined perception functions while performing high-quality communication, and improve the overall performance and service capabilities of the network. It is considered to be one of the basic core technologies in 6G wireless networks. However, most of the existing ISAC systems are based on cellular network architecture. This centralized deployment method takes the enhancement of fixed-cell coverage as the starting point, and there is a significant loss of edge performance. In addition, with the densification of 6G mobile communication networks and the complexity of network environments, the network topology is highly irregular, causing serious interference between cells and frequent cell switching problems.
[0003] The Cell-free Multiple-input Multiple-output (CFMM) system distributes a large number of access points (APs) over a wide area and connects them to the central processing unit (CPU) through fronthaul links. This distributed network architecture can effectively solve the problem of user performance loss at the cell edge by shortening the transmission path, and can significantly reduce interference between users and improve perceived service performance through collaborative processing between APs. From the perspective of engineering implementation, each AP in the CFMM is equipped with a small number of antennas, which effectively reduces the implementation complexity and cost, making it easier to implement than a centralized large-scale antenna system.
[0004] In the scenario of sudden terminal access, such as the scenario of smart transportation in the Internet of Things, a large number of users access the network at the same time, causing a sharp increase in interference, resulting in a rapid decline in perception and communication performance. Non-orthogonal multiple access (NOMA) technology can effectively solve the problems of terminal access and interference management. By using superposition coding at the transmitter and successive interference cancellation (SIC) at the receiver, NOMA supports multiple users to occupy the same time-frequency resources, thereby effectively improving the number of system connections and bandwidth efficiency. Combined with NOMA technology, the cellular-free interawareness integration (CFMM-ISAC) system can meet the needs of multiple access and improve the service quality of perception and communication. However, researchers have proposed a CFMM system design and ISAC system design based on NOMA, but the design of the related NOMA-assisted CFMM-ISAC system is still in a blank stage. Therefore, it is necessary to design a NOMA-assisted CFMM-ISAC system to improve the access capability and service performance of the CFMM-ISAC system and solve the problem of performance loss of the CFMM-ISAC system in the scenario of sudden user access.
[0005] At present, researchers have proposed CFMM system design and ISAC system design based on NOMA, but the design of related NOMA-assisted CFMM-ISAC system is still in a blank stage. In addition, in the existing ISAC system design or CFMM system design based on NOMA, researchers only considered the system beamforming design, without considering user clustering or pairing strategies. Usually, user clustering or pairing strategies play an important role in NOMA-assisted systems and directly affect system performance. The decoding order of SIC depends on the channel conditions of the users. Due to interference from other users in the same user cluster, the performance of users with weaker channel gain will decrease. And the error propagation characteristics of SIC may cause the performance of users with better channel conditions to deteriorate. Therefore, NOMA technology can perform at its best when two users are paired. Summary of the invention
[0006] In view of the above problems, the present invention proposes a NOMA-assisted de-cellularized synaesthesia integrated system beamforming method to adapt to the actual CFMM-ISAC system characteristics and requirements. The core of the present invention is to consider combining NOMA technology with CFMM-ISAC technology, give play to the advantages of NOMA technology in user access, interference management and resource utilization, and consider the NOMA user pairing strategy at the same time, propose an adaptive NOMA user pairing and beamforming design, and introduce a user pairing factor to characterize the NOMA pairing relationship of any user in the user cluster. The joint beamforming design maximizes the system communication rate while ensuring the perceived service quality. The present invention effectively improves the user access capability and synaesthesia integrated service performance in the CFMM-ISAC system through the joint design of NOMA user pairing and beamforming in the system, meets the needs of sudden terminal access scenarios, and has significant application value.
[0007] The present invention provides a beamforming design method in a NOMA-assisted de-cellularized synaesthesia integrated system, which fully utilizes the advantages of NOMA technology and de-cellularized network architecture, and at the same time considers the influence of user inter-cluster interference and serial interference cancellation error propagation characteristics on NOMA technology. Through the collaborative design of NOMA user pairing and system beamforming, the system access capability and service performance are significantly improved. The method can be applied to the design of transmitter products of the de-cellularized synaesthesia integrated system, and is specifically implemented by the following technical scheme.
[0008] The present invention provides a NOMA-assisted decellularized synaesthesia integrated system beamforming method, comprising the following steps:
[0009] Construct a downlink communication model of a NOMA-assisted decellularized synaesthesia integrated system, wherein the downlink communication module includes a transmission model of multiple communication data streams passing through an access point AP for downlink communication, an expected receiving signal of the communication data stream, a target receiving signal, and a downlink achievable rate;
[0010] Construct a radar perception model of the NOMA-assisted decellularized synaesthesia integrated system, in which the radar target is regarded as a virtual communication user. The radar perception model includes the radar target's azimuth information and radar perception energy.
[0011] An optimization problem of joint NOMA user pairing and beamforming is constructed according to the downlink communication model and the radar perception model, wherein the communication reachable rate of the de-cellularized synaesthesia integrated system is maximized to obtain multiple constraints of the optimization problem, and user pairing is achieved according to the multiple constraints;
[0012] The optimization problem is solved based on a semidefinite relaxation algorithm to obtain a target optimization problem, and initial point construction and beamforming calibration are performed on the target optimization problem to complete the beamforming of the NOMA-assisted de-cellularized synaesthesia integrated system.
[0013] As a preferred embodiment of the above technical solution, a downlink communication model of a NOMA-assisted decellularized synaesthesia integrated system is constructed, including:
[0014] The M superimposed communication data streams are transmitted through all APs for downlink communication and radar target perception at the same time. m The launch model is expressed as:
[0015]
[0016] Among them, s c,u Indicates user UE c,u The data stream signal has normalized energy, m∈M, w m,c,u ∈C N Indicates AP m To UE c,u The beamforming vector, C N represents N-dimensional complex vector space;
[0017] Preset P max is the maximum transmit power of the AP, then the transmitted signal x m The transmit power constraints that need to be met are: Where E{·} represents the expected value operation, ||·|| 2 Represents the vector two-norm square operation;
[0018] Preset definition h m,c,u ∈C N ~CN(0,R m,c,u ) is AP m With UE c,u The channel between them, CN(0,R m,c,u ) means that the mean is 0 and the covariance is R mk The complex Gaussian distribution, R m,c,u ∈ N×N represents the spatial correlation matrix, β m,c,u =tr(R m,c,u ) / N is the large-scale fading coefficient, tr(·) represents the matrix trace operation;
[0019] Preset UE c,u and UE c,u' is the selected NOMA user pair, then UE c,u Need to obtain in UE c,u and UE c,u' The target signal at the UE is used for SIC operation. c,u The received signal at can be expressed as:
[0020]
[0021] Among them, n c,u ~CN(0,σ 2 ) indicates UE c,u Additive Gaussian white noise at [·] H represents the matrix conjugate transpose operation, Desired signal represents the desired signal, Intra-cluster interference represents the interference within the user cluster, Inter-cluster interference represents the interference between user clusters, and Noise represents the noise. c,u' UE c,u The received signal is expressed as:
[0022]
[0023]
[0024] Among them, n c,u' ~CN(0,σ 2 ) indicates UE c,u' Additive white Gaussian noise at .
[0025] As a preferred embodiment of the above technical solution, variable α is introduced c,u,u' represents user cluster K c UE c,u and UE c,u' The NOMA pairing factor of is expressed as:
[0026]
[0027] Among them, Condition 1 indicates UE c,u and UE c,u' is selected as a NOMA user pair, and UE c,u' For weak users in UE c,u Decoding is performed before;
[0028] Preset user cluster K c All users in the system have been pre-scaled according to the large-scale fading factor β m,c,u Sort in descending order, then UE c,1 For the strongest users, UE c,U For the weakest user;
[0029] Determine the user's location on the UE based on the received signal expression and multiple presets c,u and UE c,u' The signal-to-interference-plus-noise ratio SINR at the position is calculated as follows:
[0030]
[0031]
[0032] in, α∈{α c,u,u'} and w∈{w c,u}, C∈C,UE c,u The SINR is:
[0033]
[0034] in, Indicates minimum operation, UE c,u The calculation expression of the downlink achievable rate is:
[0035] R c,u (α,w)=log(1+γ c,u (α,w)) (8).
[0036] As a preferred embodiment of the above technical solution, a radar perception model of a NOMA-assisted decellularized synaesthesia integrated system is constructed, including:
[0037] The expression of the incident signal is:
[0038]
[0039] Among them, θ m Indicates the position information of each AP and radar target obtained through deep learning algorithm. Indicates AP m The launch wave control vector, θ∈{θ m}, λ and d represent the carrier wavelength and antenna spacing respectively, and satisfy d = λ / 2. Then the radar sensing energy can be calculated as:
[0040]
[0041] As a preferred embodiment of the above technical solution, an optimization problem of joint NOMA user pairing and beamforming is constructed according to the downlink communication model and the radar perception model, including:
[0042] Maximize the user's minimum SINR and construct the optimization problem as:
[0043]
[0044]
[0045]
[0046]
[0047]
[0048]
[0049]
[0050]
[0051]
[0052] in, represents the minimum power required to ensure the perception performance, constraint (a) ensures the radar perception performance, constraint (b) ensures that the maximum transmit power meets the power constraint, and constraints (c) to (i) are the NOMA user pairing conditions; where constraints (c) and (d) represent the user pairing criteria, constraint (e) ensures that the NOMA user pairing satisfies the user channel fading order; constraints (f) to (i) ensure that each user is paired with only one user.
[0053] As a preferred embodiment of the above technical solution, solving the optimization problem based on a semidefinite relaxation algorithm to obtain an optimization problem includes:
[0054] Presets Among them, W c,u ≥0, rank(W c,u )=1, rank(·) represents the rank of the matrix, c∈C, the power constraint can be converted to:
[0055]
[0056] in, c∈C, formula (5) and formula (6) can be rewritten as:
[0057]
[0058]
[0059] and
[0060] According to the transmit power constraint of each AP, M auxiliary sparse matrices A are introduced m ∈R M×M , Matrix A m The mth diagonal element of is 1, and the other elements are 0. Then the power constraint (b) in the optimization problem (11) can be rewritten as:
[0061]
[0062] According to the transformation of formula (15), the original optimization problem (11) is equivalent to:
[0063]
[0064]
[0065] Constraints (c) to (i), formula (13), formula (14) (16)
[0066] The objective function in the optimization problem (16) is a non-smooth function. The intermediate variable x is introduced to transform the objective function into a smooth function. The corresponding expression is:
[0067]
[0068] Constraints (c) to (i), (13), (14), (a1), (b1) (17)
[0069] According to formula (17), the objective function is an affine function, the required variables α and W are coupled to each other, and the rank-one constraint and the binary characteristics of the NOMA pairing factor make the constraints of the optimization problem non-convex.
[0070] As a preferred embodiment of the above technical solution, according to the rank-one constraint and the binary characteristics of the NOMA pairing factor, the constraint of the optimization problem is made non-convex, so that the transformed optimization problem is difficult to solve directly, including:
[0071] The binary variable α is conditionally relaxed and approximated as a continuous variable, corresponding to α c,u,u' ∈{0,1} is converted to 0≤α c,u,u' ≤1,
[0072] According to the definition of user SINR, constraint (a2) can be equivalent to:
[0073]
[0074]
[0075] The right-hand side term in formula (18) Right side item Components is the variable α c,u,u” and W c,u” A bilinear function of is neither convex nor concave;
[0076] A continuous convex approximation scheme is used to solve The item is rewritten as:
[0077]
[0078]
[0079] Among them, the first item A c,u,u” is convex, and the second term B c,u,u” Non-convex, at a feasible point Based on (α c,u,u” ,W c,u” ) Performing a first-order Taylor expansion on it, we can get B c,u,u” The lower bound of , the corresponding expression is:
[0080]
[0081] Where i represents the number of iterations. The other approximate forms of the constraints of the optimization problem (11) are obtained. After continuous convex approximation and ignoring the rank-one constraints, the optimization problem can be obtained as follows:
[0082]
[0083]
[0084]
[0085]
[0086]
[0087]
[0088]
[0089] W c,u ≥0
[0090] Constraints (d) to (i) (22)
[0091] in, c∈C are all introduced intermediate variables, D c,u 、Ec,u,u' and F c,u,u' Respectively expressed as:
[0092]
[0093]
[0094]
[0095] Among them, formula (22) is a convex semidefinite programming problem, which is effectively solved using CVX convex optimization tools.
[0096] As a preferred embodiment of the above technical solution, the initial point construction includes:
[0097] Initialize the feasible point (α (0) ,W (0) ,x (0) ,ζ (0) ,η (0) ,ξ (0) ), the initial point construction algorithm is used to introduce an error parameter variable δ≥0 to construct a feasible problem for the optimization problem (22). The corresponding expression is:
[0098]
[0099]
[0100]
[0101]
[0102]
[0103]
[0104]
[0105] W c,u ≥0
[0106] Constraints (d) to (i) (26)
[0107] Among them, m∈M, u, u'∈K c ,c∈C,δ represents the difference between the constraint conditions and the original optimization problem (22). The optimization problem (26) is a convex problem and is solved using the CVX convex optimization tool.
[0108] As a preferred embodiment of the above technical solution, the beamforming calibration includes:
[0109] Get the converged binary variable α, when α c,u',u= 0, γ in formula (14) c,u,u' (α,W) will become infinite, given the value of α UE c,u The SINR can be expressed as:
[0110]
[0111] The corresponding target optimization problem can be constructed as:
[0112]
[0113] The optimization problem (28) is solved by the bisection method.
[0114] As a preferred embodiment of the above technical solution, the rank-one constraint is ignored when solving the original optimization problem, and the obtained solution W needs to be further processed. The eigenvalue decomposition algorithm is used to construct a rank-one constraint matrix from the matrix W of arbitrary rank
[0115] The present invention provides a NOMA-assisted decellularized synaesthesia integrated system beamforming method, by constructing a downlink communication model of the NOMA-assisted decellularized synaesthesia integrated system, constructing a radar perception model of the NOMA-assisted decellularized synaesthesia integrated system, constructing an optimization problem of joint NOMA user pairing and beamforming according to the downlink communication model and the radar perception model, solving the optimization problem based on a semidefinite relaxation algorithm to obtain a target optimization problem, performing initial point construction and beamforming calibration on the target optimization problem respectively to complete the beamforming of the NOMA-assisted decellularized synaesthesia integrated system, and comparing the results of the present invention with those of the present invention. The existing de-cellularized synaesthesia integrated system architecture, the present invention introduces a non-orthogonal access method, through the superposition coding at the transmitting end and the serial interference removal at the receiving end, the system access performance and resource utilization can be improved; at the same time, based on the NOMA technical characteristics, the present invention considers the user pairing strategy, and performs a joint NOMA user pairing and beamforming design, and flexibly performs user pairing according to different scenario requirements to improve the system synaesthesia integrated service performance; in addition, the present invention adopts semi-definite relaxation and continuous convex approximation methods to ensure system performance while maintaining controllable computational complexity, meeting the requirements of sudden terminal access scenarios of the de-cellularized synaesthesia integrated system. BRIEF DESCRIPTION OF THE DRAWINGS
[0116] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0117] Figure 1 Schematic diagram of the NOMA-assisted CFMM-ISAC system provided by the present invention;
[0118] Figure 2 A flow chart of the NOMA-assisted de-cellularized synaesthesia integrated system beamforming method provided by the present invention;
[0119] Figure 3 A simplified flow chart of the NOMA-assisted CFMM-ISAC system beamforming method provided by the present invention;
[0120] Figure 4 A flow chart of the joint NOMA user pairing and beamforming method provided by the present invention;
[0121] Figure 5 A user pairing diagram in the NOMA-assisted CFMM-ISAC system provided by the present invention;
[0122] Figure 6 The CDF diagram of the minimum achievable rate of downlink users provided by the present invention; DETAILED DESCRIPTION
[0123] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0124] The purpose of the present invention is to address the deficiencies in the prior art and provide a NOMA-assisted beamforming method in a de-cellularized synaesthesia integrated system, and to combine the NOMA user pairing design to improve the number of system access users and synaesthesia integrated service performance to meet the needs of sudden terminal access scenarios.
[0125] See also Figure 2 and Figure 3 The present invention provides a NOMA-assisted decellularized synaesthesia integrated system beamforming method, comprising the following steps:
[0126] S1: Construct a downlink communication model of a NOMA-assisted decellularized synaesthesia integrated system, wherein the downlink communication module includes a transmission model of multiple communication data streams passing through an access point AP for downlink communication, an expected receiving signal of the communication data stream, a target receiving signal, and a downlink achievable rate;
[0127] S2: Construct a radar perception model of the NOMA-assisted decellularized synaesthesia integrated system, in which the radar target is regarded as a virtual communication user. The radar perception model includes the radar target's azimuth information and radar perception energy.
[0128] S3: constructing an optimization problem of joint NOMA user pairing and beamforming according to the downlink communication model and the radar perception model, wherein the communication reachable rate of the de-cellularized synaesthesia integrated system is maximized to obtain multiple constraints of the optimization problem, and user pairing is achieved according to the multiple constraints;
[0129] S4: Solving the optimization problem based on the semidefinite relaxation algorithm to obtain the target optimization problem, performing initial point construction and beamforming calibration on the target optimization problem to complete the beamforming of the NOMA-assisted de-cellularized synaesthesia integrated system.
[0130] In this embodiment, a NOMA-assisted CFMM-ISAC system is considered to perform downlink communication and radar perception simultaneously, such as Figure 1 As shown. In a certain area, M APs are randomly distributed. Each AP is equipped with N uniformly linearly arranged antenna arrays (which can be directly expanded to uniformly distributed planar arrays) and is connected to a CPU, serving K single-antenna users and 1 radar sensing target at the same time. Without loss of generality, the K users are divided into C user clusters, and each user cluster contains U users. For ease of representation, the AP set, user set, and user cluster set are represented as: M = {1,2,...,M}, K = {1,2,...,K}, and C = {1,2,...,C}, respectively. Let K c represents the set of users in the cth user cluster, and for c, and When K=U c∈ C K c , Let UE c,u and AP m They represent the u-th user and the m-th AP in the c-th user cluster respectively.
[0131] It should be noted that the downlink communication model of the NOMA-assisted decellularized synaesthesia integrated system includes:
[0132] The M superimposed communication data streams are transmitted through all APs for downlink communication and radar target perception at the same time. m The launch model is expressed as:
[0133]
[0134] Among them, s c,u Indicates user UE c,u The data stream signal has normalized energy, m∈M, w m,c,u ∈C N Indicates AP m To UE c,uThe beamforming vector, C N represents N-dimensional complex vector space;
[0135] Preset P max is the maximum transmit power of the AP, then the transmitted signal x m The transmit power constraints that need to be met are: Where E{·} represents the expected value operation, ||·|| 2 Represents the vector two-norm square operation;
[0136] Preset definition h m,c,u ∈C N ~CN(0,R m,c,u ) is AP m With UE c,u The channel between them, CN(0,R m,c,u ) means that the mean is 0 and the covariance is R mk The complex Gaussian distribution, R m,c,u ∈ N×N represents the spatial correlation matrix, β m,c,u =tr(R m,c,u ) / N is the large-scale fading coefficient, tr(·) represents the matrix trace operation;
[0137] Preset UE c,u and UE c,u' is the selected NOMA user pair, then UE c,u Need to obtain in UE c,u and UE c,u' The target signal at the UE is used for SIC operation. c,u The received signal at can be expressed as:
[0138]
[0139] Among them, n c,u ~CN(0,σ 2 ) indicates UE c,u Additive Gaussian white noise at [·] H represents the matrix conjugate transpose operation, Desired signal represents the desired signal, Intra-cluster interference represents the interference within the user cluster, Inter-cluster interference represents the interference between user clusters, and Noise represents the noise. c,u' UE c,u The received signal is expressed as:
[0140]
[0141]
[0142] Among them, n c,u' ~CN(0,σ 2 ) indicates UE c,u' Additive white Gaussian noise at .
[0143] Specifically, the variable α is introduced c,u,u' represents user cluster K c UE c,u and UE c,u' The NOMA pairing factor of is expressed as:
[0144]
[0145] Among them, Condition 1 indicates UE c,u and UE c,u' is selected as a NOMA user pair, and UE c,u' For weak users in UE c,u Decoding is performed before;
[0146] Preset user cluster K c All users in the system have been pre-scaled according to the large-scale fading factor β m,c,u Sort in descending order, then UE c,1 For the strongest users, UE c,U For the weakest user;
[0147] Determine the user's location on the UE based on the received signal expression and multiple presets c,u and UE c,u' The signal-to-interference-plus-noise ratio SINR at the position is calculated as follows:
[0148]
[0149]
[0150] in, α∈{α c,u,u'} and w∈{w c,u}, c∈C,UE c,u The SINR is:
[0151]
[0152] in, Indicates minimum operation, UE c,uThe calculation expression of the downlink achievable rate is:
[0153] R c,u (α,w)=log(1+γ c,u (α,w)) (8).
[0154] Specifically, a radar perception model of a NOMA-assisted decellularized synaesthesia integrated system is constructed, including:
[0155] The expression of the incident signal is:
[0156]
[0157] Among them, θ m Indicates the position information of each AP and radar target obtained through deep learning algorithm. Indicates AP m The launch wave control vector, θ∈{θ m}, λ and d represent the carrier wavelength and antenna spacing respectively, and satisfy d = λ / 2. Then the radar sensing energy can be calculated as:
[0158]
[0159] Specifically, an optimization problem of joint NOMA user pairing and beamforming is constructed according to the downlink communication model and the radar perception model, including:
[0160] Maximize the user's minimum SINR and construct the optimization problem as:
[0161]
[0162]
[0163]
[0164]
[0165]
[0166]
[0167]
[0168]
[0169]
[0170] in, represents the minimum power required to ensure the perception performance, constraint (a) ensures the radar perception performance, constraint (b) ensures that the maximum transmit power meets the power constraint, and constraints (c) to (i) are the NOMA user pairing conditions; where constraints (c) and (d) represent the user pairing criteria, constraint (e) ensures that the NOMA user pairing satisfies the user channel fading order; constraints (f) to (i) ensure that each user is paired with only one user.
[0171] Specifically, the optimization problem is solved based on a semidefinite relaxation algorithm to obtain an optimization problem, including:
[0172] Presets Among them, W c,u ≥0, rank(W c,u )=1, rank(·) represents the rank of the matrix, c∈C, the power constraint can be converted to:
[0173]
[0174] in, c∈C, formula (5) and formula (6) can be rewritten as:
[0175]
[0176]
[0177] and
[0178] According to the transmit power constraint of each AP, M auxiliary sparse matrices A are introduced m ∈R M×M , Matrix A m The mth diagonal element of is 1, and the other elements are 0. Then the power constraint (b) in the optimization problem (11) can be rewritten as:
[0179]
[0180] According to the transformation of formula (15), the original optimization problem (11) is equivalent to:
[0181]
[0182]
[0183] Constraints (c) to (i), formula (13), formula (14) (16)
[0184] The objective function in the optimization problem (16) is a non-smooth function. The intermediate variable x is introduced to transform the objective function into a smooth function. The corresponding expression is:
[0185]
[0186] Constraints (c) to (i), (13), (14), (a1), (b1) (17)
[0187] According to formula (17), the objective function is an affine function, the required variables α and W are coupled to each other, and the rank-one constraint and the binary characteristics of the NOMA pairing factor make the constraints of the optimization problem non-convex.
[0188] Specifically, according to the rank-one constraint and the binary characteristics of the NOMA pairing factor, the constraints of the optimization problem are made non-convex constraints, making it difficult to solve the transformed optimization problem directly, including:
[0189] The binary variable α is conditionally relaxed and approximated as a continuous variable, corresponding to α c,u,u' ∈{0,1} is converted to 0≤α c,u,u' ≤1,
[0190] According to the definition of user SINR, constraint (a2) can be equivalent to:
[0191]
[0192]
[0193] The right-hand side term in formula (18) Right side item Components is the variable α c,u,u” and W c,u” A bilinear function of is neither convex nor concave;
[0194] A continuous convex approximation scheme is used to solve The item is rewritten as:
[0195]
[0196] Among them, the first item A c,u,u” is convex, and the second term B c,u,u” Non-convex, at a feasible point Based on (α c,u,u” ,W c,u” ) Performing a first-order Taylor expansion on it, we can get Bc,u,u” The lower bound of , the corresponding expression is:
[0197]
[0198] Where i represents the number of iterations. The other approximate forms of the constraints of the optimization problem (11) are obtained. After continuous convex approximation and ignoring the rank-one constraints, the optimization problem can be obtained as follows:
[0199]
[0200]
[0201]
[0202]
[0203]
[0204]
[0205]
[0206]
[0207] W c,u ≥0
[0208] Constraints (d) to (i) (22)
[0209] in, c∈C are all introduced intermediate variables, D c,u 、E c,u,u' and F c,u,u' Respectively expressed as:
[0210]
[0211]
[0212]
[0213] Wherein, formula (22) is a convex semidefinite programming problem, which is efficiently solved using CVX convex optimization tools. Table 1 summarizes the process of solving the optimization problem (22). In row 9, each α (i) The elements in the set are incremented by 0.5 to make the obtained optimization variables satisfy the binary characteristics.
[0214] Table 1 Joint user pairing and beamforming algorithm
[0215]
[0216] Specifically, the initial point construction includes:
[0217] According to Table 1, it is necessary to initialize the feasible point (α (0) ,W (0) ,x (0) ,ζ (0) ,η (0) ,ξ (0) ), however, it is difficult to randomly generate feasible points. To this end, we propose an initial point construction algorithm, that is, to introduce an error parameter variable δ ≥ 0 into the optimization problem (22) to construct a feasible problem. The corresponding expression is:
[0218]
[0219]
[0220]
[0221]
[0222]
[0223]
[0224]
[0225] W c,u ≥0
[0226] Constraints (d) to (i) (26)
[0227] Among them, m∈M, u, u'∈K c ,c∈C,δ represents the difference between the constraint conditions and the original optimization problem (22). The optimization problem (26) is a convex problem and is solved using the CVX convex optimization tool.
[0228] Specifically, beamforming calibration includes:
[0229] According to Table 1, we can obtain the convergent binary variable α, when α c,u',u = 0, γ in formula (14) c,u,u' (α,W) will become infinite, given the value of α UE c,u The SINR can be expressed as:
[0230]
[0231] The corresponding target optimization problem can be constructed as:
[0232]
[0233] The optimization problem (28) is solved by the binary method. It is summarized in Table 2. In each iteration, φ is the binary variable that changes in each iteration, and ε is the predefined error threshold value due to adjusting the accuracy of the algorithm.
[0234] Table 2 Beamforming calibration algorithm
[0235]
[0236]
[0237] Specifically, see Figure 4 Since the rank-one constraint is ignored when solving the original optimization problem, the solution W needs to be further processed. The eigenvalue decomposition algorithm is used to construct a rank-one constraint matrix from the matrix W of arbitrary rank The specific process of the eigenvalue decomposition algorithm is as follows: 1) Initialize the variables in the optimization problem (27) and the maximum number of iterations error threshold; 2) Use the CVX convex optimization tool to solve the optimization problem (27); 3) Determine δ (i) Is it lower than the predefined error threshold? If so, execute step 4); if not, execute step 2); 4) The solution of optimization problem (27) is used as the initial point of optimization problem (23) to calculate the NOMA user pairing factor and beamforming matrix; 5) Determine whether the algorithm converges or reaches the predefined maximum number of iterations. If so, execute step 6); if not, execute step 4); 6) Solve the optimization problem (29) based on the obtained NOMA user pairing factor; 7) Determine whether the bisection stop suite is satisfied. If so, execute step 8); if not, execute step 6); 8) Perform eigenvalue decomposition on the beamforming matrix and construct a rank-one constraint matrix; 9) Output the optimal NOMA user pairing factor and beamforming vector.
[0238] Compared with the existing de-cellularized synaesthesia integrated system architecture, the present invention introduces a non-orthogonal access method, which can improve the system access performance and resource utilization by superimposing coding at the transmitting end and removing serial interference at the receiving end; at the same time, based on the NOMA technical characteristics, the present invention considers the user pairing strategy, performs joint NOMA user pairing and beamforming design, and flexibly pairs users according to different scenario requirements to improve the system synaesthesia integrated service performance; in addition, the present invention adopts semidefinite relaxation and continuous convex approximation methods to ensure system performance while maintaining controllable computational complexity, meeting the requirements of sudden terminal access scenarios of the de-cellularized synaesthesia integrated system.
[0239] Consider that all APs, communication users, and radar targets are randomly distributed in a 200m×200m square area. Channel h m,c,u The large-scale fading coefficient β m,c,u It is expressed as:
[0240] β m,c,u [dB] = -30.5-36.7log 10 (d m,c,u )+K m,c,u (29)
[0241] Among them, d m,c,u Indicates AP m and UE c,u Distance, K m,c,u ~CN(0,4 2 ) represents the shadow fading coefficient, u∈K c ,c∈C. The maximum transmit power of each AP is set to 20dBm, and the noise power at each user is set to -70dBm. The minimum perceived power requirement is set to The error threshold for user-adjusted algorithm accuracy is set to δ min =10 -6 , ε=10 -6 The other system parameters are set as: K=12, C=3, M=8 and N=4.
[0242] Figure 5 represents the optimal user pairing result obtained according to the proposed solution. Cluster1 / Cluster2 / Cluster3 represents user cluster 1 / user cluster 2 / user cluster 3, and Target represents the perception target. It can be seen that {UE 2,1 ,UE 2,4}、{UE 2,2 ,UE 2,3}、{UE 3,3 ,UE 3,4} constitutes the optimal NOMA user pair, while there is no user pairing in user cluster 1, which shows the superior flexibility of the proposed scheme. In a given scenario, non-orthogonal and orthogonal schemes can be flexibly applied according to needs, rather than only selecting orthogonal schemes or NOMA schemes with given user pairing strategies, such as the far user pairing strategy (the strategy of pairing two users with the largest difference in channel fading coefficients).
[0243] Figure 6The Cumulative Distribution Function (CDF) graph of the minimum achievable rate of downlink communication users is given. Where Proposed represents the proposed scheme. In order to verify the effectiveness of the proposed algorithm, the performance of other access methods is also provided, namely the traditional Orthogonal Multiple Access (OMA) scheme and the NOMA scheme with different user pairing strategies (random user pairing, far user pairing and near user pairing, respectively represented as NOMA-pair, NOMA-far and NOMA-close). The simulation results show that the proposed scheme has the optimal achievable rate. Since the proposed scheme can flexibly pair users according to system requirements and calculate the optimal beamforming at the same time, its performance is better than the existing schemes. In addition, the proposed scheme achieves a performance gain of about 10% compared with the OMA scheme under the condition of 5% outage capacity. Compared with the NOMA-close scheme, a performance gain of about 30% is achieved.
[0244] It should be understood that the present invention proposes a NOMA-assisted CFMM-ISAC system architecture, combines the advantages of NOMA technology to improve the number of system access users and resource utilization, and performs joint NOMA user pairing and beamforming design based on the NOMA technical characteristics. It adopts semidefinite relaxation, continuous convex approximation and other algorithms to achieve the improvement of synaesthesia integrated service performance with low computational complexity. It proposes initial point construction and beam calibration methods for the problem of difficulty in generating initial points in the joint NOMA user pairing and beamforming method and the binary constraint problem of the NOMA user pairing factor, so as to further improve system performance and realize reliable signal transmission of the system.
[0245] In all examples shown and described herein, any specific values should be interpreted as merely exemplary and not as limiting, and thus other examples of the exemplary embodiments may have different values.
[0246] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.
[0247] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.
Claims
1. A NOMA-assisted decellularized synaesthesia integrated system beamforming method, characterized in that: The following steps are involved: Construct a downlink communication model of a NOMA-assisted decellularized synaesthesia integrated system, wherein the downlink communication module includes a transmission model of multiple communication data streams passing through an access point AP for downlink communication, an expected receiving signal of the communication data stream, a target receiving signal, and a downlink achievable rate; Construct a radar perception model of the NOMA-assisted decellularized synaesthesia integrated system, in which the radar target is regarded as a virtual communication user. The radar perception model includes the radar target's azimuth information and radar perception energy. An optimization problem of joint NOMA user pairing and beamforming is constructed according to the downlink communication model and the radar perception model, wherein the communication reachable rate of the de-cellularized synaesthesia integrated system is maximized to obtain multiple constraints of the optimization problem, and user pairing is achieved according to the multiple constraints; The optimization problem is solved based on a semidefinite relaxation algorithm to obtain a target optimization problem, and initial point construction and beamforming calibration are performed on the target optimization problem to complete the beamforming of the NOMA-assisted de-cellularized synaesthesia integrated system.
2. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 1 is characterized in that: Construct a downlink communication model of a NOMA-assisted decellularized synaesthesia integrated system, including: The M superimposed communication data streams are transmitted through all APs for downlink communication and radar target perception at the same time. m The launch model is expressed as: Among them, s c,u Indicates user UE c,u The data stream signal has normalized energy, m∈M, w m,c,u ∈C N Indicates AP m To UE c,u The beamforming vector, C N represents N-dimensional complex vector space; Preset P max is the maximum transmit power of the AP, then the transmitted signal x m The transmit power constraints that need to be met are: Among them, E{·} represents the expected value operation, ‖‖·|| 2 Represents the vector two-norm square operation; Preset definition h m,c,u ∈C N ~CN(0,R m,c,u ) is AP m With UE c,u The channel between them, CN(0, R m,c,u ) means that the mean is 0 and the covariance is R mk The complex Gaussian distribution, R m,c,u ∈ N×N represents the spatial correlation matrix, β m,c,u =tr(R m,c,u ) / N is the large-scale fading coefficient, tr(·) represents the matrix trace operation; Preset UE c,u and UE c,u′ is the selected NOMA user pair, then UE c,u Need to obtain in UE c,u and UE c,u′ The target signal at the UE is used for SIC operation. c,u The received signal at can be expressed as: Among them, n c,u ~CN(0,σ 2 ) indicates UE c,u Additive Gaussian white noise at [·] H represents the matrix conjugate transpose operation, Desired signal represents the desired signal, Intra-cluster interference represents the interference within the user cluster, Inter-cluster interference represents the interference between user clusters, and NOise represents the noise. c,u′ UE c,u The received signal is expressed as: Among them, n c,u′ ~CN(0,σ 2 ) indicates UE c,u′ Additive white Gaussian noise at .
3. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 2 is characterized in that: Introducing variable α c,u,u′ Indicates the UE in user cluster Kc c,u and UE c,u′ The NOMA pairing factor of is expressed as: Among them, Condition 1 indicates UE c,u and UE c,u′ is selected as a NOMA user pair, and UE c,u′ For weak users in UE c,u Decoding is performed before; Preset user cluster K c All users in the system have been pre-scaled according to the large-scale fading factor Sort in descending order, then UE c,1 For the strongest users, UE c,U For the weakest user; Determine the user's location on the UE based on the received signal expression and multiple presets c,u and UE c,u′ The signal-to-interference-plus-noise ratio SINR at the position is calculated as follows: in, α∈{α c,u,u′ } and w∈{w c,u }, c∈C,UE c,u The SINR is: in, Indicates minimum operation, UE c,u The calculation expression of the downlink achievable rate is: R c,u (α, w)=log(1+γ c,u (a,w)) (8).
4. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 1 is characterized in that: Construct a radar perception model for a NOMA-assisted decellularized synaesthesia integrated system, including: The expression of the incident signal is: Among them, θ m Indicates the position information of each AP and radar target obtained through deep learning algorithm. Indicates AP m The launch wave control vector, λ and d represent the carrier wavelength and antenna spacing respectively, and satisfy d = λ / 2, then the radar sensing energy can be calculated as:
5. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 1 is characterized in that: An optimization problem of joint NOMA user pairing and beamforming is constructed according to the downlink communication model and the radar perception model, including: Maximize the user's minimum SINR and construct the optimization problem as: in, represents the minimum power required to ensure the perception performance, constraint (a) ensures the radar perception performance, constraint (b) ensures that the maximum transmit power meets the power constraint, and constraints (c) to (i) are the NOMA user pairing conditions; where constraints (c) and (d) represent the user pairing criteria, constraint (e) ensures that the NOMA user pairing satisfies the user channel fading order; constraints (f) to (i) ensure that each user is paired with only one user.
6. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 5 is characterized in that: Solving the optimization problem based on the semidefinite relaxation algorithm to obtain the optimization problem includes: Presets Among them, W c,u ≥0, rank(W c,u )=1, rank(·) represents the rank of the matrix, Then the power constraint can be converted to: in, Formula (5) and formula (6) can be rewritten as: and According to the transmit power constraint of each AP, M auxiliary sparse matrices are introduced Matrix A m The mth diagonal element of is 1, and the other elements are 0. Then the power constraint (b) in the optimization problem (11) can be rewritten as: According to the transformation of formula (15), the original optimization problem (11) is equivalent to: Constraints (c) to (i), formula (13), formula (14) (16) The objective function in the optimization problem (16) is a non-smooth function. The intermediate variable x is introduced to transform the objective function into a smooth function. The corresponding expression is: Constraints (c)~(i), (13), (14), (a1), (b1) (17) According to formula (17), the objective function is an affine function, the required variables α and W are coupled to each other, and the rank-one constraint and the binary characteristics of the NOMA pairing factor make the constraints of the optimization problem non-convex constraints.
7. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 6 is characterized in that: According to the rank-one constraint and the binary characteristics of the NOMA pairing factor, the constraints of the optimization problem are made non-convex, making it difficult to solve the transformed optimization problem directly, including: The binary variable α is conditionally relaxed and approximated as a continuous variable, corresponding to α c,u,u′ ∈{0, 1} is converted to According to the definition of user SINR, constraint (a2) can be equivalent to: The right-hand side term in formula (18) Right side item Components is the variable α c,u,u” and W c,u” A bilinear function of is neither convex nor concave; A continuous convex approximation scheme is used to solve The item is rewritten as: Among them, the first item A c,u,u” is convex, and the second term B c,u,u” Non-convex, at a feasible point Based on (α c,u,u” , W c,u” ) Performing a first-order Taylor expansion on it, we can get B c,u,u” The lower bound of , the corresponding expression is: Where i represents the number of iterations. The other approximate forms of the constraints of the optimization problem (11) are obtained. After continuous convex approximation and ignoring the rank-one constraints, the optimization problem can be obtained as follows: in, are introduced intermediate variables, D c,u 、E c,u,u′ and F c,u,u′ Respectively expressed as: Among them, formula (22) is a convex semidefinite programming problem, which is effectively solved using CVX convex optimization tools.
8. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 7 is characterized in that: The initial point construction includes: Initialize feasible points The initial point construction algorithm is used to introduce an error parameter variable δ≥0 to the optimization problem (22) to construct a feasible problem. The corresponding expression is: Among them, m∈M, u, u′∈K c , c∈C, δ represents the difference between the constraints and the original optimization problem (22). The optimization problem (26) is a convex problem and is solved using the CVX convex optimization tool.
9. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 7, characterized in that: Beamforming calibration includes: Get the converged binary variable α, when α c,u′,u = 0, γ in formula (14) c,u,u′ (α, W) will become infinite, given the value of α UE c,u The SINR can be expressed as: The corresponding target optimization problem can be constructed as: The optimization problem (28) is solved by the bisection method.
10. The NOMA-assisted decellularized synaesthesia integrated system beamforming method according to claim 9, characterized in that: When solving the original optimization problem, the rank-one constraint is ignored. The solution W obtained needs to be further processed. The eigenvalue decomposition algorithm is used to construct a rank-one constraint matrix from the matrix W of arbitrary rank.
Citation Information
Cited By
Node mode selection and beam forming method in cooperative sensing integrated scene
CN119483668A
Node mode selection and beamforming method in collaborative synaesthesia integration scenario
CN119483668B
Communication and perception integration method and device based on partial non-orthogonal multiple access
CN119893691A
Artificial noise-assisted honeycomb-removing and sensing-eliminating integrated beam forming method
CN121585217A
