A method for mode selection and transceiver design in NOMA-assisted ISAC system

Abstract

This paper discloses a method for mode selection and transceiver design for a non-orthogonal multiple access (NOMA)-assisted integrated sensing and communication (ISAC) in a network-assisted full-duplex, cell-free massive multiple-input, multiple-output (MIMO) system. The method includes the following steps: transforming the joint mode selection and transceiver design problem of the NOMA-ISAC in the cell-free massive MIMO system into an optimization problem for maximizing the radar-sensing signal-to-interference-plus-noise ratio (SINR), decomposing the problem into two subproblems. The optimal operating mode, precoding vector, and uplink power are obtained using a simulated annealing algorithm and a continuous convex approximation algorithm, respectively. Simulation results show that the method can effectively improve the system's perception performance while meeting the communication user's rate constraints.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and in particular to a mode selection and transceiver design method in a NOMA-assisted ISAC system. Background Art

[0002] Integrated sensing and communication (ISAC) is one of the six major application scenarios of the sixth-generation mobile communication system. ISAC enables devices to perceive their surroundings while communicating, playing an important role in promoting emerging industries including the Internet of Things and autonomous driving.

[0003] However, integrating communication and perception functions on the same device also brings strong interference problems, which poses a challenge to the actual deployment of ISAC. To solve this problem, more advanced interference management technologies, such as non-orthogonal multiple access (NOMA), are used to assist the ISAC system. At the same time, in order to give full play to the potential coordinated perception capabilities of multiple base stations, we integrate NOMA-ISAC into a network-assisted full-duplex cell-free massive MIMO system for research. The network-assisted full-duplex (NAFD) that has emerged in recent years can complete data transmission and reception on the same time-frequency resource block. Compared with the traditional TDD / FDD mode, it can double the spectrum efficiency of the mobile communication link. At the same time, compared with the CCFD mode, it can effectively alleviate the self-interference problem caused by full-duplex, thereby effectively improving the system performance.

[0004] Implementing NOMA-ISAC in network-assisted full-duplex cell-free massive MIMO systems still presents many challenges. A reasonable RAU operating mode and transceiver design can effectively improve system performance. Therefore, joint optimization of the operating mode and transceiver design is crucial. Summary of the Invention

[0005] The purpose of the present invention is to provide a mode selection and transceiver design method in a NOMA-assisted ISAC system to effectively improve system performance.

[0006] To achieve this object, the present invention adopts the following technical solutions:

[0007] A method for mode selection and transceiver design in a NOMA-assisted ISAC system includes the following steps:

[0008] Step 1: Establish an optimal receiver expression for the NOMA-assisted ISAC in a network-assisted full-duplex, cell-free massive MIMO system, thereby deriving an expression for the system-perceived SINR. Then, a joint optimization of duplex mode selection, downlink precoding, and uplink power allocation is performed to maximize the system-perceived SINR and modeled as a mathematical optimization problem. Since the optimization problem is non-convex, a block coordinate descent method is used to decompose the optimization problem into two sub-problems: mode selection and transceiver design. The optimal solution is obtained using a simulated annealing algorithm and a continuous convex approximation algorithm, respectively.

[0009] Step 2: When the number of iterations of the simulated annealing algorithm reaches the upper limit, the optimal working mode is output; when the number of iterations of the continuous convex approximation algorithm reaches the upper limit, the transceiver design scheme is output.

[0010] Preferably, the optimization problem model for maximizing the system perceived SINR is:

[0011]

[0012] Where w={w k |0≤k≤K D} and p={p j |1≤j≤K U} respectively represent downlink precoding w k and the set of uplink power, T m =diag[0 (m-1)N ,1 N ,0 (M-m)N ], M represents the number of remote antenna units, N represents the number of antennas, K U Indicates the number of uplink users, K D represents the number of downlink users, T represents the total number of radars, q D ∈{0, 1} M×1 Select the downlink RAU working mode vector, q D (m) indicates the downlink RAU working mode corresponding to the mth remote antenna unit. If it is 1, it means the mth remote antenna unit is in downlink transmission mode, otherwise it is in uplink reception mode. j represents the uplink power of user j, γ D,k represents the SINR of downlink user k, which is obtained by decoding the signal s k When γ , downlink user k regards the signals from other downlink users and all uplink users as interference; U,j represents the SINR of uplink user j, which is obtained by CPU analyzing the signal d from uplink user j. jWhen decoding, the signals from the uplink users with higher priority have been decoded and completely eliminated from the received signal, and the signals from the uplink users with lower priority and the signals of all downlink users are treated as interference; γ D,min , γ U,min 、P U,max 、P D,max They represent the minimum SINR required by downlink users, the minimum SINR required by uplink users, the maximum transmit power of uplink users, and the maximum transmit power of downlink remote antenna units respectively; γ s,t represents the perceived SINR corresponding to the radar target t.

[0013] Preferably, in step 1, the block coordinate descent method is used to decompose the optimization problem into two sub-problems: mode selection and transceiver design, and the optimal solution is obtained based on the simulated annealing algorithm and the continuous convex approximation algorithm respectively. The specific steps include:

[0014] Step 1.1: Initialize q D , randomly generate a 0-1 vector, where the i-th element represents the working mode of the i-th RAU. If it is 1, it works in downlink transmission mode, and if it is 0, it works in uplink reception mode;

[0015] Step 1.2: Under a given operating mode, find the optimal downlink precoding and uplink power based on continuous convex approximation, and obtain the corresponding perceived SINR.

[0016] Step 1.3: Set the maximum number of iterations L of the simulated annealing algorithm out , the control factor θ is a pre-set parameter used to control the population convergence speed. If the current number of iterations does not reach the set upper limit, the evolution operation is performed; during the evolution process, q D The mth element q in D (m) is changed to 1-q D (m), by solving the optimization problem (1), we can obtain the optimal downlink precoding and uplink power corresponding to the new duplex mode, as well as the corresponding perceived SINR; the perceived SINR before and after the modification are respectively denoted as and

[0017] Step 1.4: If Then q D (m) is changed to 1-q D (m), if Assume q D (m) The probability of still having P is changed to 1-q D (m), that is, if q D If (m) is 1, change it to 0; otherwise, change it to 1. The expression of P is

[0018]

[0019] When the current number of iterations reaches the set upper limit, go to step 2.

[0020] Preferably, step 1.2 specifically includes:

[0021] Define the matrix And ignoring the rank constraint, we get The SINR of downlink user k is expressed as

[0022]

[0023] in, represents the effective channel vector of downlink user k to all RAUs, represents the downlink RAU mode selection matrix, and the uplink RAU mode selection matrix is h D，k represents the channel vector from downlink user k to all RAUs, h IUI，j，k represents the channel between downlink user k and uplink user j, Indicates the noise power at the downlink user;

[0024] The downlink RAU power constraint in the optimization problem shown in equation (1) is rewritten as

[0025]

[0026] On this basis, under a given duplex mode, the sub-problems of downlink precoding and uplink power optimization are remodeled as

[0027]

[0028] Based on the continuous convex approximation algorithm, the non-convex problem shown in the above formula (5) is transformed into the following convex problem for solution

[0029]

[0030] Among them, n s ={n s，t |1≤t≤T} and m s ={m s,t |1≤t≤T} is the temporary variable n introduced s,t and m s,t A collection of Represents m s,t and n s,t The value at the nth iteration, and Respectively represent Ψ t and Φj The value at the nth iteration, represents the echo signal matrix reflected by radar target t, h e，D，t represents the channel from radar target t to all downlink RAUs, Uplink RAU mode selection matrix for providing radar target perception services, represents the effective channel vector from uplink user j to all RAU antennas, h U，j represents the channel vector from uplink user j to all RAU antennas, represents the uplink noise power, η is the radar cross section, E represents the perturbation matrix; function g(.), f t (.) and k j (.) are defined as

[0031] in, h e，U，t represents the channel from radar target t to all uplink RAUs; the optimal downlink precoding and uplink power in a given working mode are obtained by multiple iterations of equation (6), and the corresponding perceived SINR is obtained.

[0032] Preferably, in step 2, check whether the current iteration number i has reached the maximum number of iterations L out , if i<L out , then return to step 1, otherwise, output the current optimal duplex mode selection and transceiver design scheme.

[0033] Beneficial Effects: The method of the present invention transforms the RAU duplex mode and transceiver design of the NOMA-ISAC in a network-assisted full-duplex non-cellular massive MIMO system into vectors to be optimized, transforming the joint optimization problem of mode selection and transceiver design into an optimization problem of maximizing the system-perceived SINR. An algorithm based on simulated annealing and continuous convex approximation is proposed to solve this problem. Compared with the simulated annealing-based mode selection method, the continuous convex approximation-based transceiver design method, and the MRT precoding and random mode selection methods, the method of the present invention can effectively improve system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is an algorithm flow chart of a specific embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of a network-assisted full-duplex cell-free massive MIMO system according to a specific embodiment of the present invention;

[0036] Figure 3 This is a performance comparison diagram of the method according to the specific embodiment of the present invention and the MRT precoding and random mode selection methods. DETAILED DESCRIPTION

[0037] The present invention will be further described below with reference to the accompanying drawings.

[0038] First, let's make the following explanation: Consider a typical single-cell network-assisted full-duplex non-cellular massive MIMO system with M remote antenna units (RAUs). All RAUs are distributed in the cell according to a Poisson distribution and connected to the central processing unit (CPU) through a low-latency backhaul link. Each RAU is equipped with N antennas. There are K U uplink users and K D The total number of users and radars is K and T respectively. The channel model and signal transmission model of the system are briefly introduced below.

[0039] The channel vector from downlink user k to all RAU antennas is defined as but

[0040]

[0041] Where, represents the large-scale fading matrix, λ m,k is the large-scale fading factor from RAU m to downlink user k, represents the small-scale fading vector.

[0042] Similarly, the channel vector from uplink user j to all antennas on the RAU is defined as The channel from uplink user j to downlink user k is The self-interference matrix between RAUs is: The echo signal matrix reflected by radar target t is H e，t It can be expressed as

[0043]

[0044] In the network-assisted full-duplex cell-free massive MIMO system, for the convenience of expression, the binary vector q is defined as U ∈{0, 1} M×1 and q D ∈{0, 1} M×1 As the uplink and downlink RAU working mode selection vectors respectively, specifically, if RAUi works in the uplink receiving mode, then q U (i) = 1 and q D (i) = 0; otherwise, if RAUi is working in downlink transmission mode, then q D (i) = 1 and qU (i)=0, it is easy to get q U +q D =1 M×1 In addition, the RAU mode selection matrix of the system is further defined as and

[0045] For downlink transmission, the signals sent by M RAUs It can be expressed as

[0046]

[0047] in, and They represent the symbols sent to downlink user u and for target sensing, and Represent the corresponding precoding vectors respectively. Therefore, the signal received by downlink user k is It can be expressed as

[0048]

[0049] in, represents the signal sent by uplink user j, then the power of uplink user j can be expressed as p j =E{|d j | 2}. Additionally, is additive Gaussian white noise, satisfying

[0050] Similarly, for uplink transmission, the signal received by all RAUs It can be expressed as

[0051]

[0052] in, is independent and identically distributed additive Gaussian white noise, satisfying

[0053] Since supporting NOMA requires more complex hardware, it is often difficult to implement for single-antenna users. Therefore, it is assumed that the NOMA strategy is only used in uplink communication to assist the ISAC system. Without loss of generality, it is assumed that uplink users are arranged in increasing order of channel quality, that is, For downlink user k, after decoding s k When , the signals from other downlink users and all uplink users are regarded as interference, and the corresponding communication SINR can be obtained as

[0054]

[0055] in, represents the effective channel vector from downlink user k to all RAUs. When the CPU receives the signal d from uplink user j, j When decoding, the signals from the uplink users with higher priority have already been decoded and are completely eliminated from the received signal. Define the receive vector To decode d j , then the corresponding SINR is

[0056]

[0057] in, represents the effective channel vector from uplink user j to all RAU antennas, represents the effective interference channel caused by the radar target echo and the downlink transmission. In addition, the covariance matrix of the downlink ISAC signal can be written as After decoding and eliminating all signals from uplink users, define To capture the reflected signal of radar target t. Additionally, it is assumed that only the L uplink RAUs closest to the sensing area provide sensing services for the radar target. Definition and As the corresponding binary uplink RAU selection vector and matrix, it is necessary to satisfy On this basis, the perception SINR corresponding to the radar target t is:

[0058]

[0059] in, and The corresponding optimal filters are

[0060]

[0061] in, Similarly, we can get E represents the perturbation matrix. Substituting the above optimal filters into the corresponding SINR expressions, we can get

[0062]

[0063] Duplex mode selection, downlink precoding and uplink power allocation have a direct impact on the performance of NOMA-assisted ISAC in network-assisted full-duplex non-cellular massive MIMO system. Reasonable duplex mode selection, downlink precoding and uplink power allocation are crucial to improving system performance. To this end, the present invention optimizes duplex mode selection, downlink precoding and uplink power allocation to maximize the system's perceived SINR while meeting the communication user rate constraint. This optimization problem can be abstracted into the following mathematical problem

[0064]

[0065] Where w={w k |0≤k≤K D} and p={p j |1≤j≤K U} represent the set of downlink precoding and uplink power respectively. m =diag[0 (m-1)N , 1 N , 0 (M-m)N ].

[0066] Based on the analysis of communication and radar perception SINR in the above system, the working mode selection, downlink precoding and uplink power allocation method based on simulated annealing algorithm and continuous convex approximation algorithm provided by the present invention can be divided into the following steps:

[0067] Step 1: Initialize q D , randomly generate a 0-1 vector, the i-th element represents the working mode of the i-th RAU, 1 means it works in downlink transmission mode, and 0 means it works in uplink reception mode.

[0068] Step 2: Under a given working mode, the optimal downlink precoding and uplink power are solved based on continuous convex approximation. Inspired by the semi-positive relaxation algorithm, the matrix is ​​defined as And ignore the rank constraint. Then we can get The SINR of downlink user k can be expressed as

[0069]

[0070] in, In addition, the downlink RAU power constraint in the optimization problem shown in Equation (13) can be rewritten as

[0071]

[0072] On this basis, under a given duplex mode, the sub-problems of downlink precoding and uplink power optimization can be remodeled as

[0073]

[0074] The optimization problem shown in formula (16) is difficult to solve because the objective function and some constraints are non-convex. Based on the continuous convex approximation algorithm, the above non-convex problem is transformed into the following convex problem for solution

[0075]

[0076] Among them, n s ={n s，t|1≤t≤T} and m s ={m s，t |1≤t≤T} is the set of temporary variables introduced. Functions g(.), f t (.) and k j (.) are defined as

[0077] and Respectively represent Ψ t and Φ j The value at the nth iteration. By iteratively calculating formula (17) multiple times, the optimal downlink precoding and uplink power in a given working mode are obtained, and the corresponding perceived SINR is obtained.

[0078] Step 3: Set the maximum iteration number L of the simulated annealing algorithm out , control factor θ, if the current number of iterations does not reach the upper limit, the evolution operation is performed. During the evolution process, according to step 2, q D The mth element q in D (m) is changed to 1-q D The optimal downlink precoding and uplink power corresponding to the duplex mode of (m), and the corresponding perceived SINR. The perceived SINR before and after the modification is respectively denoted as and

[0079] Step 4: If Then q D (m) is changed to 1-q D (m), if Different from the greedy strategy, assuming q D (m) The probability of still having P is changed to 1-q D (m), the expression of P is

[0080]

[0081] In this way, we can avoid falling into the local optimal solution. When the current number of iterations reaches the set upper limit, go to step 5;

[0082] Step 5: Output the current optimal duplex mode selection and transceiver design solution.

[0083] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for mode selection and transceiver design in a NOMA-assisted ISAC system, characterized by: The following steps are involved: Step 1: Establish an optimal receiver expression for the NOMA-assisted ISAC in a network-assisted full-duplex, cell-free massive MIMO system, thereby deriving an expression for the system-perceived SINR. Then, a joint optimization of duplex mode selection, downlink precoding, and uplink power allocation is performed to maximize the system-perceived SINR and modeled as a mathematical optimization problem. Since the optimization problem is non-convex, a block coordinate descent method is used to decompose the optimization problem into two sub-problems: mode selection and transceiver design. The optimal solution is obtained using a simulated annealing algorithm and a continuous convex approximation algorithm, respectively. Step 2: When the number of iterations of the simulated annealing algorithm reaches the upper limit, the optimal working mode is output; when the number of iterations of the continuous convex approximation algorithm reaches the upper limit, the transceiver design scheme is output; The optimization problem model for maximizing the system perceived SINR is: Where w={w k |0≤k≤K D } and p={p j |1≤j≤K U } respectively represent downlink precoding w k and the set of uplink power, T m =diag[0 (m-1)N , 1 N , 0 (M-m)N ], M represents the number of remote antenna units, N represents the number of antennas, K u Indicates the number of uplink users, K D represents the number of downlink users, T represents the total number of radars, q D ∈{0,1} M×1 Select the downlink RAU working mode vector, q D (m) indicates the downlink RAU working mode corresponding to the mth remote antenna unit. If it is 1, it means the mth remote antenna unit is in downlink transmission mode, otherwise it is in uplink reception mode. j represents the uplink power of user j, γ D,k represents the SINR of downlink user k, which is obtained by decoding the signal s k When γ , downlink user k regards the signals from other downlink users and all uplink users as interference; U,j represents the SINR of uplink user j, which is obtained by CPU analyzing the signal d from uplink user j. j When decoding, the signals from the uplink users with higher priority have been decoded and completely eliminated from the received signal, and the signals from the uplink users with lower priority and the signals of all downlink users are treated as interference; γ D,min , γ U,min 、P U,max 、P D,max They represent the minimum SINR required by downlink users, the minimum SINR required by uplink users, the maximum transmit power of uplink users, and the maximum transmit power of downlink remote antenna units respectively; γ s,t represents the perceived SINR corresponding to the radar target t; In step 1, the block coordinate descent method is used to decompose the optimization problem into two sub-problems: mode selection and transceiver design. The optimal solution is obtained based on the simulated annealing algorithm and the continuous convex approximation algorithm, respectively. The specific steps include: Step 1.1: Initialize q D , randomly generate a 0-1 vector, where the i-th element represents the working mode of the i-th RAU. If it is 1, it works in downlink transmission mode, and if it is 0, it works in uplink reception mode; Step 1.2: Under a given operating mode, find the optimal downlink precoding and uplink power based on continuous convex approximation, and obtain the corresponding perceived SINR. Step 1.3: Set the maximum number of iterations L of the simulated annealing algorithm out , the control factor θ is a pre-set parameter used to control the population convergence speed. If the current number of iterations does not reach the set upper limit, the evolution operation is performed; during the evolution process, q D The mth element q in D (m) is changed to 1-q D (m), the optimal downlink precoding and uplink power corresponding to the new duplex mode and the corresponding perceived SINR are obtained by solving the optimization problem; the perceived SINR before and after the modification are respectively recorded as and Step 1.4: If Then q D (m) is changed to 1-q D (m), if Assume q D (m) The probability of still having P is changed to 1-q D (m), that is, if q D If (m) is 1, change it to 0; otherwise, change it to 1. The expression of P is When the current number of iterations reaches the set upper limit, go to step 2.

2. The method for mode selection and transceiver design in a NOMA-assisted ISAC system according to claim 1, wherein: Step 1.2 specifically includes: Define the matrix 0≤k≤K D And ignoring the rank constraint, we get The SINR of downlink user k is expressed as in, represents the effective channel vector of downlink user k to all RAUs, represents the downlink RAU mode selection matrix, and the uplink RAU mode selection matrix is h D，k represents the channel vector from downlink user k to all RAUs, h IUI，j，k represents the channel between downlink user k and uplink user j, Indicates the noise power at the downlink user; The downlink RAU power constraint in the optimization problem shown in equation (1) is rewritten as On this basis, under a given duplex mode, the sub-problems of downlink precoding and uplink power optimization are remodeled as Based on the continuous convex approximation algorithm, the non-convex problem shown in the above formula (5) is transformed into the following convex problem for solution Among them, n s ={n s，t |1≤t≤T} and m s ={m s，t |1≤t≤T} is the temporary variable n introduced s,t and m s,t A collection of Represents m s,t and n s,t The value at the nth iteration, C represents the effective interference channel caused by radar target echo and downlink transmission; and Respectively represent Ψ t and Φ j The value at the nth iteration, represents the echo signal matrix reflected by radar target t, h e，D，t represents the channel from radar target t to all downlink RAUs, Uplink RAU mode selection matrix for providing radar target perception services, represents the effective channel vector from uplink user j to all RAU antennas, h U，j represents the channel vector from uplink user j to all RAU antennas, Indicates the uplink noise power, H IRI represents the self-interference matrix between RAUs, η is the radar cross section, E represents the disturbance matrix; the functions g(·), f t (·) and k j (·) are defined as in, h e，U，t represents the channel from radar target t to all uplink RAUs; the optimal downlink precoding and uplink power in a given working mode are obtained by multiple iterations of equation (6), and the corresponding perceived SINR is obtained.

3. The method for mode selection and transceiver design in a NOMA-assisted ISAC system according to claim 1, wherein: In step 2, check whether the current iteration number i has reached the maximum number of iterations L out , if i <L out , then return to step 1, otherwise, output the current optimal duplex mode selection and transceiver design scheme.

Images