Integrated base station calculation unloading method based on IRS assistance

Abstract

The invention relates to the technical field of radar and communication, and particularly discloses an integrated base station calculation unloading method based on IRS assistance, which comprises the following steps of: unloading received data to a server for centralized processing by means of an intelligent reflection surface (IRS) technology in a base station near-end high-performance edge server environment, and constructing a radar detection and communication uplink joint receiving system; and then, under receiver power constraint, frequency band energy constraint and IRS constant modulus constraint, with minimization of total energy consumption of the system as a radar point target, constructing a calculation unloading problem model so as to perform synchronous optimization on receiving vector design. And finally, decomposing the calculation unloading problem model, and respectively solving a radar phase matrix and a communication receiving vector IRS phase matrix. According to the method, the overall performance of each functional module is improved, the adaptive capacity and the anti-interference capacity of the system in an actual complex environment are enhanced, and the global optimal balance of the communication rate, the sensing precision, the time delay and the energy efficiency is realized.

Description

Technical Field

[0001] The present invention relates to the field of radar and communication technology, and in particular to an integrated base station calculation offloading method based on IRS assistance. Background Art

[0002] Integrated sensing and computing deeply integrates communication, perception, and computing functions. Through resource sharing and collaborative optimization, it achieves comprehensive improvements in spectrum efficiency, energy efficiency, and multi-dimensional resource utilization. As 6G networks evolve towards integrated sensing and computing, radar and communication systems face unprecedented challenges within a spectrum-sharing architecture. Radar and communication systems compete for limited spectrum resources, significantly increasing the complexity of spectrum resource allocation, especially in the time-frequency-space-energy multi-dimensional resource domain. Furthermore, integrated nodes often lack efficient computing capabilities, posing a computing challenge to processing data received by communication radars.

[0003] Traditional separate optimization methods typically treat radar and communication systems as independent entities and optimize their designs separately, ignoring their mutual influence and coupling. However, in the integrated 6G network, radar and communication systems are closely connected across multiple dimensions, including spectrum, time, space, and energy. Optimization of either will impact the other. Therefore, a novel optimization method is needed that comprehensively considers the interaction between radar and communication systems and enables coordinated scheduling and dynamic optimization of multidimensional resources. Furthermore, the mutual interference between radar echoes and communication uplink signals, as well as the stringent requirements of edge computing tasks in energy-constrained environments, create a core contradiction between high-precision radar detection, high-quality communication services, and system energy consumption. This constrains overall system performance and poses new challenges for resource coordinated scheduling and dynamic optimization. Researchers have proposed a multifunctional beamforming design framework that integrates perception, communication, and computing. By jointly optimizing the Cramer-Rao lower bound (CRB), communication signal-to-interference-plus-noise ratio (SINR), and computation rate, efficient resource allocation is achieved. The proposed semidefinite relaxation (SDR) algorithm significantly improves radar point target estimation performance while ensuring a rank-one solution. Other researchers have proposed a mobile edge computing network optimization solution for 6G ultra-reliable, low-latency communications. By jointly designing perception, communication, and computing resources, they minimize the number of services deployed on edge servers and the end-to-end latency of communication users while meeting strict latency and reliability requirements. A two-stage optimization approach is used, combining short-term task offloading and bandwidth allocation optimization with long-term service deployment strategies. However, existing methods require preset fixed weights or priorities and lack dynamic adjustment mechanisms, making it difficult to achieve a global optimal balance between communication rate, perception accuracy, latency, and energy efficiency in complex scenarios. The joint optimization framework involves high-dimensional non-convex problems, and existing algorithms based on SDR and other methods have high computational complexity, making them difficult to scale to large-scale networks. Summary of the Invention

[0004] The present invention provides an integrated base station computing offloading method based on IRS assistance, which solves the technical problem that existing methods are difficult to achieve a global optimal balance among communication rate, perception accuracy, latency and energy efficiency in complex scenarios.

[0005] To solve the above technical problems, the present invention provides an integrated base station calculation offloading method based on IRS assistance, comprising the steps of:

[0006] S1. Construct a radar detection and communication uplink joint receiving system, which includes an integrated base station, a server, N communication users, a radar point target, and an intelligent reflective surface (IRS). The radar point target sends data to the integrated base station via the intelligent reflective surface when blocked by obstacles. The communication user sends data directly to the integrated base station and also sends data to the integrated base station via the intelligent reflective surface. The integrated base station is also interfered by K clutter signals. The integrated base station offloads communication computing and radar computing tasks to the server.

[0007] S2. Taking the minimization of the total system energy consumption as the radar point goal, a computational offloading problem model is constructed to jointly optimize the communication computation offloading variables, radar computation offloading variables, and IRS reflection coefficient variables while satisfying the receiver power constraints of the integrated base station, the frequency band energy constraints, and the IRS constant modulus constraints.

[0008] S3. Solve the computation offloading problem model and obtain the optimal solution of the communication computation offloading variable, the radar computation offloading variable, and the IRS reflection coefficient variable.

[0009] Furthermore, step S2 specifically includes the steps of:

[0010] Construct an energy consumption minimization objective function to minimize the total energy consumption of the system by optimizing the communication computing offloading variables and the radar computing offloading variables;

[0011] By adding the IRS reflection coefficient variable, the energy consumption minimization objective function is transformed into a weighted sum rate maximization objective function that maximizes the weighted sum of the radar data offloading rate and the communication data offloading rate.

[0012] Determine the constraints, including receiver power constraints, frequency band energy constraints, and IRS constant modulus constraints;

[0013] Based on the idea of ​​solving the Rayleigh quotient problem, the radar data offloading rate function and the communication data offloading rate function in the weighted sum rate maximization objective function are simplified to obtain the optimized overall objective function.

[0014] A computational offloading problem model is obtained, which aims to achieve the optimization of the overall objective function and jointly optimizes the communication computational offloading variables, radar computational offloading variables, and IRS reflection coefficient variables under the constraints.

[0015] Furthermore, the receiver power constraint is that the norm of the power vector of each communication receive filter is 1, and the norm of the power vector of the radar receive filter is 1; the frequency band energy constraint is that within each restricted sub-band of the radar, the energy of the radar receiving end must be less than the radar receiving energy threshold, and within the specified communication restricted sub-band, the energy of the communication receiving end must be less than the communication receiving energy threshold; the IRS constant modulus constraint is that the modulus of the reflection coefficient of each reflection unit is 1.

[0016] Furthermore, step S3 specifically includes the steps of:

[0017] S31, transforming the computation offloading problem model into a radar optimization subproblem that only optimizes radar computation offloading variables, a communication optimization subproblem that only optimizes communication computation offloading variables, and a reflection coefficient optimization subproblem that only optimizes IRS reflection coefficient variables;

[0018] S32. Iteratively solve the radar optimization subproblem, the communication optimization subproblem, and the reflection coefficient optimization subproblem.

[0019] Furthermore, in step S31, by fixing the communication calculation offloading variable and the IRS reflection coefficient variable, the constants in the calculation offloading problem model are eliminated, and the properties of the matrix determinant are utilized, with the help of the first-order Taylor expansion, and the equivalent real-valued transformation is introduced to perform an equivalent transformation on the objective function to obtain the radar optimization subproblem.

[0020] Furthermore, in the iterative solution process of step S32, the Lagrange alternating direction multiplier method is used to solve the radar optimization subproblem.

[0021] Furthermore, in step S31, by fixing the radar calculation offloading variable and the IRS reflection coefficient variable, retaining the corresponding constraints and objective function, and using the block coordinate descent method to separate and solve, with the help of the first-order Taylor expansion, an equivalent real-valued transformation is introduced to perform an equivalent transformation on the objective function, and the communication optimization subproblem is obtained.

[0022] Furthermore, in the iterative solution process of step S32, the Lagrange alternating direction multiplier method is used to solve the communication optimization subproblem.

[0023] Furthermore, in step S31, by fixing the radar calculation offloading variables and the communication calculation offloading variables, retaining the corresponding constraints and objective function, and using continuous convex approximation, first-order Taylor expansion, and matrix transformation to perform equivalent transformation on the objective function, the IRS reflection coefficient optimization subproblem is obtained.

[0024] Furthermore, the objective function of the computational offloading problem model is to maximize B r is the bandwidth of the radar system, SINR r is the received signal-to-interference-and-noise ratio of the receiving filter designed for the radar target point, B c,n is the bandwidth of communication user n, SINR c,n is the received signal-to-interference-and-noise ratio of the receiving filter designed for communication user n, λ c,n is the offload task ratio of communication user n, λ r is the ratio of radar unloading tasks,

[0025] The present invention provides an integrated base station computation offloading method based on IRS assistance. First, in a high-performance edge server environment near the base station, with the help of intelligent reflecting surface (IRS) technology, the received data is offloaded to the server for centralized processing, and a radar detection and communication uplink joint reception system is constructed. Then, under the constraints of receiver power, frequency band energy, and IRS constant modulus, a computation offloading problem model is constructed with the radar point target of minimizing the total system energy consumption, so as to simultaneously optimize the receiving vector design. Finally, the computation offloading problem model is decomposed, and the radar and communication receiving vectors are respectively relaxed using first-order Taylor expansion, and then alternately solved using the alternating direction penalty (ADPM) algorithm. The IRS phase matrix is ​​replaced by the upper bound problem of the second-order function, and then simplified using the block coordinate descent (BCD) algorithm for element-by-element solution. The introduction of IRS can optimize the channel environment, improve communication quality, and effectively improve the joint performance of the system. This method achieves overall performance improvement of each functional module, enhances the system's adaptability and anti-interference ability in actual complex environments, and achieves a global optimal balance between communication rate, perception accuracy, latency, and energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 1 is an architecture diagram of a radar detection and communication uplink joint receiving system provided by an embodiment of the present invention;

[0027] Figure 2 is a graph showing changes in performance indicator values ​​versus the number of iterations provided by an embodiment of the present invention;

[0028] Figure 3 This is a comparison curve diagram of the impact of IRS on system performance provided by an embodiment of the present invention;

[0029] Figure 4 is a graph showing a change in the objective function value as the CNR changes, provided by an embodiment of the present invention;

[0030] Figure 5is a curve diagram showing the change of the communication objective function along with the radar interference-to-noise ratio provided by an embodiment of the present invention;

[0031] Figure 6 4 is a curve diagram showing the change of the radar target function along with the communication interference-noise ratio provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following describes the embodiments of the present invention in detail with reference to the accompanying drawings. The embodiments are provided for illustrative purposes only and are not to be construed as limiting the present invention. The accompanying drawings are provided for reference and illustration only and do not constitute a limitation on the scope of protection of the present invention. Many changes may be made to the present invention without departing from the spirit and scope of the present invention.

[0033] An embodiment of the present invention provides an integrated base station calculation offloading method based on IRS assistance, comprising the steps of:

[0034] S1. Build a radar detection and communication uplink joint reception system, which includes an integrated base station (base station node), a server, N communication users, one radar point target, and an intelligent reflecting surface (IRS). The radar point target sends data to the integrated base station via the intelligent reflecting surface when blocked by obstacles. The communication user sends data directly to the integrated base station and also sends data to the integrated base station via the intelligent reflecting surface. The integrated base station is also interfered by K clutter signals. The integrated base station offloads communication computing and radar computing tasks to the server.

[0035] S2. Taking the minimization of total system energy consumption as the radar point goal, a computational offloading problem model is constructed to jointly optimize the communication computational offloading variables and the radar computational offloading variables while satisfying the receiver power constraints of the integrated base station, the frequency band energy constraints, and the IRS constant modulus constraints.

[0036] S3. Solve the computation offloading problem model and obtain the optimal solutions for the communication computation offloading variables and the radar computation offloading variables.

[0037] The present invention first uses intelligent reflecting surface (IRS) technology to offload received data to the server for centralized processing in a high-performance edge server environment near the base station, and constructs a radar detection and communication uplink joint receiving system. Then, under the constraints of receiver power, frequency band energy, and IRS constant modulus, a computational offloading problem model is constructed with the radar point target of minimizing the total energy consumption of the system to optimize the receiving vector design. Finally, the computational offloading problem model is solved to obtain the offloading distribution of radar data and communication data. The introduction of IRS can optimize the channel environment, improve communication quality, and effectively improve the joint performance of the system. This method achieves overall performance improvement of each functional module, enhances the system's adaptability and anti-interference ability in actual complex environments, and achieves a global optimal balance between communication rate, perception accuracy, latency, and energy efficiency.

[0038] The following describes each step in more detail.

[0039] (1) Step S1: Constructing a system model

[0040] In the receiving scenario of a uniform linear array base station, considering that the base station is equipped with M receiving / transmitting antennas, its tasks include not only receiving uplink data transmission from single-antenna communication users, but also detecting the echo signals of radar point targets. Specifically, there are N communication users and a single radar point target, but in actual applications, the radar echo signal is often affected by K clutter interferences, which puts higher requirements on signal processing and radar point target detection. In order to solve the problem of difficulty in direct-view detection of radar point targets due to factors such as building obstruction, the present invention further introduces intelligent reconfigurable surfaces to assist radar point target detection, and uses its beamforming characteristics to improve the service quality of communication users. Assume that the IRS has L independent reflection units, which can be used to finely control the signal in a multipath environment, thereby enhancing the energy of the radar point target signal and suppressing interference, thereby improving the overall detection performance and communication quality of the system. The specific scenario diagram is as follows: Figure 1 shown.

[0041] In theoretical modeling, it is assumed that the received signal is known. For each communication user n, the data pulse set transmitted in the uplink is recorded as Where P represents the set of pulse numbers, represents a complex set. At the same time, for the radar point target echo signal, it is recorded as These two types of signals together constitute the base station's received signal. The introduction of IRS adds an additional degree of freedom to the signal transmission channel, allowing the signal's phase and amplitude to be more flexibly controlled, thereby achieving effective separation and joint detection of multiple radar point target signals. The received signal model is constructed as follows:

[0042]

[0043] Among them, the first term y r It is the radar part receiving signal, including radar detection target signal and clutter reflection signal. is the receiving and transmitting steering matrix of the radar detection target signal, is the path loss coefficient of the radar target signal, and They are radar point targets with respect to the radar target direction The receiving steering vector and the reflected steering vector, is the channel matrix from the smart reflective surface to the communication user n, is the diagonal matrix associated with the smart reflective surface, defined as Consistent with the definition of uniform linear array, is the reflection coefficient of the lth reflection unit, and diag{} is a diagonal matrix. is the receiving and transmitting steering matrix of the clutter reflection signal, is the path loss coefficient of active clutter interference, and They are respectively the radar point target with respect to the clutter interference direction The receive and reflect steering vectors are consistent with the definition of a uniform linear array. The superscripts T and H denote the matrix transpose and conjugate transpose, respectively.

[0044] The second term y c is the communication user uplink signal, where is the channel vector from communication user n to the base station, is the channel vector from the smart reflective surface to the communication user n, is the path loss coefficient of the communication user signal. n It is a set of data pulses transmitted by communication user n on uplink.

[0045] The third item n v is the noise term, which is additive white Gaussian noise with power δ 2 .

[0046] For radar function, a normalized receiving filter is introduced at the receiving end, and its receiving signal-to-interference-noise ratio is:

[0047]

[0048] f r is the weight vector of the radar receive filter.

[0049] For the communication function, a corresponding receiving filter is designed for each communication user n, and its receiving signal-to-interference-noise ratio is:

[0050]

[0051] f c,n is the receiving filter vector of the nth communication user, is the expectation operator, which represents the statistical mean of random variables.

[0052] (2) Step S2: Constructing a computational offloading problem model

[0053] This method takes the minimization of the total energy consumption of the system as the original objective function, and energy consumption optimization needs to consider the radar unloading energy consumption E at the same time. r and communication offloading energy consumption (E c,n represents the unloading energy consumption of communication user n), and a trade-off is made between the two to achieve the overall energy consumption E T Assume that the amount of computing task data of a single communication user n is D c,n , and its data offloading rate is B c,n is the bandwidth of communication user n; and the amount of computational data for radar detection tasks is D r , the radar data unloading rate is B r is the bandwidth of the radar system. In order to further characterize the task offloading process, the offloading task ratio λ of communication user n is introduced c,n and radar offload workload ratio λ r , which respectively measure the distribution of communication and radar computing tasks, the total energy consumption of the system can be modeled as Among them, P c and P r They are the communication radar receiver power, and satisfy the constraints The original energy consumption minimization objective function can be equivalently converted to maximizing the weighted sum rate, that is:

[0054]

[0055] In terms of constraint design, considering the limited power of the receiving filter, in order to ensure that the power of the communication and radar receiving ends is constrained and the system operates stably, the power of the communication receiving filter is set to ‖‖f c,n ‖‖=1,n=1,...,N, the power of the radar receiving filter is ‖‖f r ‖‖=1,|| || is the norm of the vector. In addition, in order to effectively suppress the interference at the receiving end, the receiver needs to be optimized and constrained in both the spatial and frequency domains. Assume that there are X invalid receiving frequency bands in the system, that is, the signals in these frequency bands should be strictly suppressed. and Represent the upper and lower limits of the xth invalid receiving frequency band, respectively. The radar receiver should set appropriate suppression weights within these frequency bands to reduce the interference effect of non-radar point target signals. Furthermore, the receiving energy of the mth radar receiving filter can be expressed as:

[0056]

[0057] f is the radar receiving filter, f r,m (p) is the receiving filter weight of the mth radar receiving antenna at the pth pulse time, f r,m is the filter weight vector of the mth radar receiving antenna, r,x,m is the energy constraint matrix of the radar receiver in the xth invalid frequency band and the mth filter.

[0058] Ξ r,x (p1, p2) is the (p1, p2)th element of the frequency domain energy constraint matrix at the radar receiver, which is:

[0059]

[0060] Among them, there are In high-priority communication mission scenarios, to ensure high-quality service for communication users, it is necessary to use signal correlation to suppress interference and to limit the spectrum of non-radar point target signals outside a specific frequency band to reduce out-of-band interference. Assume that each communication user n needs to suppress interference signals within a specific frequency band and set an invalid receiving frequency band to limit the impact of irrelevant signals. Let and They represent the upper and lower bounds of the normalized frequency interval of the vth interference suppression band of communication user n, respectively. That is, within this frequency band, the power of all non-radar point target signals should be suppressed as much as possible to avoid interfering with the communication data reception of communication user n. In this context, the reception energy of the mth communication reception filter of the nth communication user can be modeled as:

[0061]

[0062] f c,n,m (p) is the receiving filter weight of the nth communication user and the mth receiving antenna at the pth moment, f c,n,m is the filter vector of the nth communication user and the mth receiving antenna, c,n,v,m is the energy constraint matrix of the mth receiving filter of user n in the communication system in the vth interference suppression band.

[0063] Ξ c,n,v,m (p3, p4) is the (p3, p4)th element of the frequency domain energy constraint matrix at the communication receiving end, which is:

[0064]

[0065] Likewise, there are

[0066] In order to improve the performance of radar detection and communication data reception, it is necessary to reasonably constrain the reception energy of each frequency band to meet the energy control requirements, optimize spectrum utilization and reduce the impact of interference. For the M-dimensional array filter, the total reception energy of the radar receiver is defined as Similarly, the total received energy at the communication receiving end is defined as I M is an M×M unit matrix. Specifically, it is stipulated that within each restricted sub-band of the radar, the energy must be less than the radar receiving energy threshold ν r , and within the specified communication restricted sub-band, the energy must be less than the communication receiving energy threshold ν c , that is, the following constraints must be met:

[0067]

[0068] In addition, in order to enhance the system’s perception and communication performance in complex wireless environments, a passive intelligent reflective surface is introduced to assist in dynamically controlling the channel state and improving the signal propagation characteristics. Since the IRS is composed of passive reflective units, its phase control capability is limited by physical implementation, so it needs to meet the constant modulus constraint, that is, for the reflection coefficient of the lth reflective unit, || means modulo.

[0069] When jointly optimizing radar echo reception and communication reception, the system needs to solve a non-convex optimization problem. To this end, the normalization characteristics of the signal and filter are used to simplify the radar objective function, thereby reducing the solution complexity. More specifically, the radar signal processing is optimized by the following steps:

[0070]

[0071] in, H b,n =h b,n +G H Θ H h b,n , I MPis the identity matrix of size M×P, and Tr() is the trace of the matrix. This optimization problem can be reduced to a generalized Rayleigh Quotient Problem (GRQP), whose goal is to maximize the signal-to-interference-noise ratio of the system under given constraints. According to the theory of generalized eigenvalue decomposition (GEVD) and the maximum eigenvalue criterion (MEC), the optimal solution of s can be obtained. in, Next, substitute the optimal receiving vector s into the original optimization problem and deduce the numerator to simplify it to:

[0072]

[0073] Therefore, the numerator is equal to 1 as a whole, and for the denominator:

[0074]

[0075] Finally, the expression can be written as According to the matrix determinant lemma, for any matrix Q and vector x, det(I+Qxx H )=1+x H Qx, where I is the identity matrix, that is, for the radar target function, we have:

[0076]

[0077] det() is the determinant of the matrix.

[0078] Similarly, for the communication signal-to-interference-and-noise ratio expression, it can first be converted into:

[0079]

[0080] Solving this Rayleigh quotient problem, we can get the optimal x n for Among them, C c,n is the interference plus noise covariance matrix of the nth communication user, I P is a unit matrix of size P×P. Substituting into the original equation, the uplink communication signal-to-noise ratio function of communication user n can be simplified to:

[0081]

[0082] Finally, the overall objective function of system optimization can be re-expressed as:

[0083]

[0084] (3) Step S3, solving the computational unloading problem model

[0085] The optimization problem shown in Equation (16) shifts from minimizing computational offloading energy consumption to maximizing the weighted sum rate that balances radar detection and communication service quality. The objective function includes both radar and communication components. To meet system constraints, power normalization, frequency band energy constraints, and IRS constant modulus constraints are introduced. This is a non-convex optimization problem. Next, consider splitting the original problem into three subproblems and solving them iteratively.

[0086] 1) Optimize radar receiving filter

[0087] fixed Eliminating the constant, the optimization problem can be transformed into:

[0088]

[0089] The properties of the matrix determinant are used to perform an equivalent transformation on the objective function to simplify the solution process. The optimization objective is reconstructed into a new expression. Among them O r (f r ) is about optimizing variable f r function, can more clearly reflect the structural properties of the optimization problem. In addition, using C(f r ) Re-express C, which represents f r function, since C(f r )and All about the optimization variable f r It has joint convexity, so it can be linearly approximated with the help of first-order Taylor expansion to construct a convex approximation problem, thereby reducing the non-convexity and solution complexity of the problem. Assume that in the current iteration, the solution of the previous iteration is Based on the differential law of complex matrices, The Taylor expansion expression nearby is:

[0090]

[0091] in, To take the real part of the complex number. Using the idea of ​​the Majorization-Minimization (MM) substitution algorithm, since the term affecting concavity after Taylor expansion has a negative value, the original maximization problem can be converted to:

[0092]

[0093] in, Band restriction matrix Ξr,x As a symmetric positive definite matrix, it has good numerical properties and can be decomposed using the Cholesky decomposition method, that is, to find a lower triangular matrix U r,x Make Introduce an equivalent real-value transformation to convert complex-valued variables into equivalent real-valued expressions, that is, use f r,R ,Ω r,R ,κ r,R ,Ξ r,x,R To express f r ,Ω r ,κ r ,Ξ r,x The real-valued form of , so that the optimization problem can be reformulated in the real number field. Finally, it is converted into an equivalent real-valued optimization problem:

[0094]

[0095] Its Lagrangian function can be expressed as:

[0096]

[0097] Among them, {μ 1,x}, μ2 is the Lagrange multiplier vector associated with the constraint condition, {υ 1,v},υ2 represents the penalty coefficient in the augmented Lagrangian function. In the iterative framework of the alternating direction multiplier method, if the current iteration is q1, the variable update rule can be implemented by the following steps:

[0098] ① Update f r,R

[0099] Only keep the constraints and the objective function related to f r,R The original Lagrangian problem can be reformulated as follows:

[0100]

[0101] Construct its about f r,R The gradient expression of , by establishing the stationary point equation and making it equal to 0, can be analytically obtained as:

[0102]

[0103] ②Update b

[0104] By constraining and eliminating terms not related to b in the objective function, the original Lagrangian problem can be reformulated as follows:

[0105]

[0106] Its closed-form solution can be given by the following formula:

[0107]

[0108] in

[0109] ③ Update {c x}

[0110] For any x=1,...,X, the original problem can be expressed as:

[0111]

[0112] Its closed-form solution can be expressed as:

[0113]

[0114] in

[0115] ④ Update {μ 1,x} and μ2

[0116] The Lagrangian parameters are updated as follows:

[0117]

[0118] ⑤ Update {υ 1,x} and υ2

[0119] The penalty parameters are updated as follows:

[0120]

[0121] in, is a number slightly larger than 1, used to speed up the convergence of the algorithm. as well as

[0122] 2) Optimize communication receiving filter

[0123] At this stage, fixed Then focus on optimizing {f c,n}, retaining the corresponding constraints and objective function, the new optimization objective can be obtained as:

[0124]

[0125] First, for each receiving filter for user n, the N terms of the objective function and the constraints are one-to-one and only related to the optimization variable f c,n They are related to each other and independent of each other, so they can be solved separately using the block coordinate descent method, that is, for any f in the set c,n , the objective function can be converted to:

[0126]

[0127] Similar to the objective function conversion of the radar part, eliminating irrelevant terms and converting it into solving the problem in, is the value of the previous iteration, O c,n (f c,n ) is the weighted sum rate function of the communication users, and the replacement function obtained by first-order Taylor expansion is:

[0128]

[0129] in, The linear term does not affect the concavity and convexity, and the original maximization communication objective function can be further transformed into:

[0130]

[0131] in, For a positive semidefinite symmetric matrix Ξ c,n,v We can find a matrix E c,n,v satisfy At the same time, use f c,n,R ,Ω c,n,R , Ξ c,n,v,R represents f c,n ,Ω c,n , Ξ c,n,v In real-valued form, for the nth user, the final optimization objective can be expressed as:

[0132]

[0133] For the original optimization problem with constraints, the augmented Lagrangian function can be constructed by introducing dual variables. Its mathematical form is defined as:

[0134]

[0135] where {μ 3,v},μ4 and {υ 3,v},υ4 are the Lagrange multiplier vectors that are penalty parameters for balancing the objective function and the constraints. In the iterative framework of the alternating direction multiplier method, the core process of the q2th iteration can be described as follows:

[0136] ① Update f c,n,R

[0137] By eliminating the constraints and the c,n,R Irrelevant terms, the original communication optimization problem can be reformulated as the following function:

[0138]

[0139] f c,n,R Taking the derivative and setting the equation equal to 0, we can analytically obtain the solution of this iteration as:

[0140]

[0141] ②Update d

[0142] By removing the constraints and the terms in the objective function that are unrelated to d, the original Lagrangian problem can be simplified to the following form:

[0143]

[0144] Its closed-form solution can be given by the following formula:

[0145]

[0146] in,

[0147] ③ Update {e v}

[0148] For any v=1,...,V, the original problem can be expressed as:

[0149]

[0150] Its closed-form solution can be expressed as:

[0151]

[0152] in,

[0153] ④ Update {μ 3,v} and μ4

[0154] The Lagrangian parameters are updated as follows:

[0155]

[0156] ⑤ Update {υ 3,v} and υ4

[0157] The penalty parameters are updated as follows:

[0158]

[0159] Among them, 0<ι1<1, 0<ι3<1, ι2, ι4 are numbers slightly larger than 1, which are used to speed up the convergence of the algorithm.

[0160]

[0161] 3) Optimize the phase of the smart reflective surface

[0162] At this stage, fixed f r ,{f c,n}optimization Only keep and The relevant constraints are converted to the original optimization problem:

[0163]

[0164] Considering the continuous convex approximation process, according to Where M refers to any matrix, is the value of the previous iteration. The linear terms on the left and right sides of the equation maintain consistent concavity and convexity. By using this property and eliminating the constant term, the objective function can be further simplified as follows:

[0165]

[0166] in, and is the derivative of the last corresponding term between the radar target point and the communication user n. It is worth noting that the converted optimization objective function has a good convexity feature in structure, where the first term is jointly convex with respect to (C(Θ), A0(Θ)), and the second term is jointly convex with respect to (C c,n (Θ),H b,n (Θ)) is also jointly convex. In order to further simplify the optimization problem and improve the convergence of the algorithm, the first-order Taylor expansion is used to linearize the objective function and construct a feasible convex optimization subproblem. Assume The internal structure of the first term of the objective function is expanded to the value obtained in the previous iteration. The complex matrix differential theory is used to perform a local linear approximation on the function so that the new optimization problem maintains convexity in each iteration. Then, the terms related to Θ are retained, that is:

[0167]

[0168] in, and variables During the optimization process, the first term of problem (46) does not contain Θ and can be ignored, while the second term contains Θ and can be expanded and simplified to Using the commutative property of the diagonal matrix Θg=diag(g)θ, where Θ=diag(θ), it can be finally converted to:

[0169]

[0170] in It is a constant term that can be omitted.

[0171] Using the commutative property of trace, Ψ r,2 A0(Θ) can finally be converted to:

[0172]

[0173] in, T2=ΘGΨ r,2 G H . For the first term of the communication part, it can be simplified to:

[0174]

[0175] in, It is irrelevant to the variable and can be ignored.

[0176] Next, we simplify the part of the communication part containing the quartic term about Θ, and use the second-order relaxation technique to replace the quartic term of Θ with an appropriate quadratic form so that it can still closely approximate the original function while maintaining the convexity constraint.

[0177]

[0178] During the optimization iteration process, is set to the result of the previous iteration, and the matrix ss H The largest eigenvalue of the unit matrix M s Since the objective function involves multiple variables, the variables related to A0 are retained, and the first term can be ignored because its structure is a constant diagonal matrix. Similarly, the third term can also be omitted as it is a constant matrix obtained in the previous iteration, thereby reducing the computational complexity.

[0179]

[0180] Using the diagonal property of Θ, in yes The i-th row element of is the i-th row of matrix G. Similarly, we have Substituting into the original formula we have By exchanging the order of integration, we get g i is the conjugate transpose of the i-th row of matrix G, Extracted to get definition is the i-th row and j-th column element of Λ, it can be simplified to: θ HΛθ. The last term can be simplified to Tr(Ψ c,n,2 G H Θ H h i,n )=Tr(Θ H h i,n Ψ c,n,2 G H )=θ H ψ, where ψ is h i,n Ψ c,n,2 G H The vector composed of the diagonal elements of i,n Ψ c,n,2 G H ) 1,1 ,(h i,n Ψ c,n,2 G H ) 2,2 ,...,(h i,n Ψ c,n,2 G H ) LP,LP ], the expression can finally be converted to:

[0181]

[0182] in, In addition, consider Due to the existence of constraints, problem (46) is still non-convex and needs to be further simplified. Consider using the BCD algorithm to sequentially calculate each smart reflective surface element. First, for the first item, For a fixed l, the sum is divided into Related parts and Irrelevant constant term: When i=l and k=l, we have When i=l and k≠l, we have When i≠l and k=l, we have When i≠l and k≠l, it can be regarded as a constant. Therefore, there is is a constant and can be ignored. Since E is a Hermitian matrix, Can be simplified and merged Adding the linear term, we have and Finally, it can be simplified to Contains The objective function part (ignoring the constant term) can be written as in Then you can get but Because |a l|>0, in order to minimize the expression, φ l -θ l =π(mod 2π), that is, we can get θ l =φ l -π, the solution is:

[0183] The effects of the present invention are verified below.

[0184] In the system model of the present invention, a radar communication integrated ULA base station configured with an edge server is considered. Its transmit / receive array is M = 6, and the number of array elements of the linear intelligent reflective surface is L = 10. The base station detects radar point targets through the reflective surface, where the transmitted signal is assumed to be a 7-bit Barker code. At the same time, two single-antenna communication users N request communication services from the base station through the uplink, where the communication duration and radar pulse are consistent, both P = 7. In addition, it is assumed that there are K = 2 active clutter forwarding radar signals to interfere with the base station. In the spectrum restriction scenario, it is assumed that there is X = 1 radar receiving frequency band restriction area with a normalized frequency interval of [0.6, 0.7], there is V = 1 communication receiving frequency band restriction area with an interval of [0.3, 0.4], and the spectrum restriction threshold is 10 -1 , the radar communication task is considered to be equally important in the initial setting, that is, λ r =0.5,λ c,1 =λ c,2 =0.25. Assuming the noise power δ 2 is 0.01, and the communication channel loss coefficient is set to The radar channel loss coefficient is set to The active clutter channel loss coefficient is set to

[0185] The present invention first studies the influence of penalty parameters on the convergence of the optimization problem. By adjusting the penalty parameters, its influence on the convergence speed and the final optimized value can be analyzed, thereby exploring the appropriate parameter selection, so that the optimization algorithm can converge quickly and achieve a good trade-off between radar and communication performance. Figure 2 It is the curve of the performance index value of the present invention changing with the number of iterations. Figure 2It can be seen that the choice of penalty parameter has a significant impact on the convergence speed and final performance of the optimization process. When the penalty parameter is set to 500, the optimization algorithm converges quickly in the first few iterations, and the overall target value stabilizes quickly, indicating that the optimization efficiency is high under this parameter and a good balance is achieved between radar and communication performance. However, as the penalty parameter increases to 1000 and 1500, although convergence is eventually achieved, the final convergence value is slightly lower than the previous values. In addition, the convergence speed decreases significantly, especially in the initial stage, and the optimization process becomes slower, indicating that larger penalty parameters may lead to hysteresis in the constraint convergence process.

[0186] The role of intelligent reflective surfaces in joint radar communication systems was then explored, and the impact of IRS on radar detection and communication transmission was analyzed by comparing the system performance with and without IRS. Figure 3 It is a comparison curve of the impact of IRS on system performance. Figure 3 The results show that radar and communication performance are doubled when the IRS is present compared to when it is not. This shows that the IRS helps enhance the target echo by regulating the reflected signal, making radar detection more accurate and stable. It also proves that the introduction of the IRS can optimize the channel environment, improve communication quality, and effectively enhance the joint performance of the system, verifying its technical advantages in spectrum sharing scenarios.

[0187] Next, this paper explores system performance changes from two perspectives: carrier-to-noise ratio (CNR) and interference-to-noise ratio (INR). In real-world scenarios, INR is often constrained by the practical limitations of transmitter power, so this paper sets the INR research range to 0-20dB. Active jammers, on the other hand, often detect the target signal and then forward it at a similar or higher power, thus extending the CNR range to 0-30dB. Figure 4 is the curve of the objective function value changing with the change of CNR. Figure 4 As can be seen from the figure, as the CNR increases, the objective function values ​​of the various optimization algorithms show significant differentiation. Specifically, the proposed algorithm is able to maintain the highest and most stable objective function value as the CNR increases, effectively mitigating the negative impact of environmental degradation on system performance. In contrast, the alternating direction multiplier method (ADMM) scheme shows a decline in overall performance, and its robustness is slightly inferior to that of the proposed algorithm. Furthermore, the performance of the SDR algorithm further deteriorates, reflecting its lack of adaptability in high-interference environments. These comparative results verify the effectiveness of the proposed scheme in combating clutter interference. It not only improves the overall performance of each functional module but also enhances the system's adaptability in complex real-world environments.

[0188] Finally, different interference-to-noise ratios are achieved by adjusting the loss coefficients of the radar channel and the communication channel respectively. Figure 5It is a curve diagram of the communication objective function changing with the radar interference-noise ratio. Figure 6 It is a curve diagram of the radar objective function changing with the communication interference-noise ratio. Figure 5 and Figure 6 It can be seen that as the INR increases, the performance of the three optimization algorithms on communication targets and radar targets declines to varying degrees. Figure 5 By analyzing the changes in the communication objective function value as the radar interference intensity increases, it can be seen that the proposed method has the strongest anti-radar interference ability and the slowest performance degradation, while the ADMM algorithm has poor performance and the SDR algorithm has the worst performance. Figure 6 The changes in the radar objective function value with the increase of communication interference intensity are analyzed. It can be seen that when the interference intensity is not large, the performance of the three algorithms is close, and the proposed algorithm has certain advantages. However, as the interference-to-noise ratio increases, the performance of the SDR algorithm and the ADMM algorithm drops sharply, while the proposed algorithm still maintains a strong anti-interference ability, which verifies the robustness of the proposed algorithm to interference.

[0189] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. An integrated base station calculation offloading method based on IRS assistance, characterized in that: Including steps: S1. Construct a radar detection and communication uplink joint receiving system, which includes an integrated base station, a server, N communication users, a radar point target, and an intelligent reflective surface (IRS). The radar point target sends data to the integrated base station through the intelligent reflective surface when blocked by an obstacle. The communication user sends data directly to the integrated base station and also sends data to the integrated base station through the intelligent reflective surface. The integrated base station is also interfered by K clutter signals. The integrated base station offloads communication computing and radar computing tasks to the server; S2. Taking the minimization of the total system energy consumption as the radar point goal, a computational offloading problem model is constructed to jointly optimize the communication computation offloading variables, radar computation offloading variables, and IRS reflection coefficient variables while satisfying the receiver power constraints of the integrated base station, the frequency band energy constraints, and the IRS constant modulus constraints. S3. Solve the computation offloading problem model and obtain the optimal solution of the communication computation offloading variable, the radar computation offloading variable, and the IRS reflection coefficient variable.

2. The IRS-assisted integrated base station calculation offloading method according to claim 1, characterized in that: Step S2 specifically includes the following steps: Construct an energy consumption minimization objective function to minimize the total energy consumption of the system by optimizing the communication computing offloading variables and the radar computing offloading variables; By adding the IRS reflection coefficient variable, the energy consumption minimization objective function is transformed into a weighted sum rate maximization objective function that maximizes the weighted sum of the radar data offloading rate and the communication data offloading rate. Determine the constraints, including receiver power constraints, frequency band energy constraints, and IRS constant modulus constraints; Based on the idea of ​​solving the Rayleigh quotient problem, the radar data offloading rate function and the communication data offloading rate function in the weighted sum rate maximization objective function are simplified to obtain the optimized overall objective function. A computational offloading problem model is obtained, which aims to achieve the optimization of the overall objective function and jointly optimizes the communication computational offloading variables, radar computational offloading variables, and IRS reflection coefficient variables under the constraints.

3. The IRS-assisted integrated base station calculation offloading method according to claim 2, characterized in that: The receiver power constraint is that the norm of the power vector of each communication receive filter is 1, and the norm of the power vector of the radar receive filter is 1. The frequency band energy constraint is that within each restricted sub-band of the radar, the energy of the radar receiver must be less than the radar receive energy threshold, and within the specified communication restricted sub-band, the energy of the communication receiver must be less than the communication receive energy threshold. The IRS constant modulus constraint is that the modulus of the reflection coefficient of each reflection unit is 1.

4. The IRS-assisted integrated base station calculation offloading method according to claim 3, characterized in that: Step S3 specifically includes the following steps: S31, transforming the computation offloading problem model into a radar optimization subproblem that only optimizes radar computation offloading variables, a communication optimization subproblem that only optimizes communication computation offloading variables, and a reflection coefficient optimization subproblem that only optimizes IRS reflection coefficient variables; S32. Iteratively solve the radar optimization subproblem, the communication optimization subproblem, and the reflection coefficient optimization subproblem.

5. The IRS-assisted integrated base station calculation offloading method according to claim 4, characterized in that: In step S31, by fixing the communication calculation offloading variable and the IRS reflection coefficient variable, the constants in the calculation offloading problem model are eliminated, and the properties of the matrix determinant are used, with the help of the first-order Taylor expansion, and the equivalent real-valued transformation is introduced to perform an equivalent transformation on the objective function to obtain the radar optimization subproblem.

6. The IRS-assisted integrated base station calculation offloading method according to claim 5, characterized in that: In the iterative solution process of step S32, the Lagrange alternating direction multiplier method is used to solve the radar optimization subproblem.

7. The IRS-assisted integrated base station calculation offloading method according to claim 4, characterized in that: In step S31, by fixing the radar calculation offloading variable and the IRS reflection coefficient variable, retaining the corresponding constraints and objective function, and using the block coordinate descent method to separate and solve, with the help of the first-order Taylor expansion, an equivalent real-valued transformation is introduced to perform an equivalent transformation on the objective function to obtain the communication optimization subproblem.

8. The IRS-assisted integrated base station calculation offloading method according to claim 7, characterized in that: During the iterative solution process of step S32, the Lagrange alternating direction multiplier method is used to solve the communication optimization subproblem.

9. The IRS-assisted integrated base station calculation offloading method according to claim 4, characterized in that: In step S31, by fixing the radar calculation offloading variables and the communication calculation offloading variables, retaining the corresponding constraints and objective function, and using continuous convex approximation, first-order Taylor expansion, and matrix transformation to perform equivalent transformation on the objective function, the IRS reflection coefficient optimization subproblem is obtained.

10. The IRS-assisted integrated base station calculation offloading method according to any one of claims 1 to 9, characterized in that: The objective function of the computational offloading problem model is to maximize B r is the bandwidth of the radar system, SINR r is the received signal-to-interference-and-noise ratio of the receiving filter designed for the radar target point, B c,n is the bandwidth of communication user n, SINR c,n is the received signal-to-interference-and-noise ratio of the receiving filter designed for communication user n, λ c,n is the offload task ratio of communication user n, λ r is the ratio of radar unloading tasks,

Images