A sparse method for intelligent reflective surfaces in a quantum deer hunting mechanism

Abstract

This invention discloses a sparse intelligent reflector method for a quantum deer hunting mechanism. First, a signal transmission system including a base station, intelligent reflector, and receiver is modeled to obtain the signal received by the receiver. Then, a system capacity calculation method is modeled for the sparse intelligent reflector to obtain the information transmission rate of the receiver after sparsification. The quantum velocity and position of each deer in the hunting population are initialized, and it is determined whether to perform an exploration or exploitation behavior. The updated position of each deer is recorded, and the fitness value is calculated using a fitness function to update the global and local optimal positions. Finally, the global optimal position is obtained. This invention automatically optimizes the reflector unit layout using a swarm intelligence optimization algorithm to find the optimal solution for the reflector unit layout. Furthermore, based on the original algorithm, quantum encoding is combined to discretize the algorithm, proposing a quantum deer hunting method. This invention possesses better global search capabilities and higher convergence accuracy.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically relating to a smart reflective surface sparse method for a quantum deer hunting mechanism. Background Technology

[0002] Smart reflectors are considered one of the most promising technologies for the future, attracting significant attention since their inception due to their low power consumption and strong compatibility. The core component of a smart reflector is the reflecting element, which can independently modulate the phase and amplitude of the incident signal. The design and performance of the reflecting element are crucial to the overall performance of the smart reflector communication system. To fully leverage the auxiliary role of smart reflectors in communication systems, current research mainly focuses on optimizing the reflection characteristics of the reflecting element and the distributed deployment of smart reflectors. Although these methods have made some progress in improving the performance of smart reflector communication systems in recent years, a technical solution that can effectively reduce the number of reflecting elements while maintaining good performance is still lacking. Sparse smart reflector schemes are an effective method to reduce system cost and increase system capacity. They can achieve equivalent performance with the same array size and fewer excitation elements while significantly reducing cost and power consumption. Therefore, sparse design of smart reflectors is imperative.

[0003] By searching the literature, Qingqing Wu et al. published "Beamforming Optimization for Wireless Network Aided by Intelligent Reflecting Surface With Discrete Phase Shifts" (2019, Vol.68, No.3, pp.1838-1851) in IEEE Transactions on Communications. In this paper, the problem is transformed into integer linear programming by discretizing the phase of the intelligent reflector surface. The optimal solution for system capacity is obtained by using branching and binding methods. However, this method has the problems of large computational load and low computational efficiency, and the calculation result is usually not the optimal solution to the problem. In their paper "Capacity Characterization for Intelligent Reflecting Surface Aided MIMO Communication" published in *IEEE Journal on Selected Areas in Communications* (2020, Vol. 38, No. 8, pp. 1823-1838), Shuowen Zhang et al. characterized the basic capacity limit of a point-to-point MIMO communication system with a reflector-assisted multi-antenna transmitter and receiver by jointly optimizing the reflector coefficients and the MIMO transmit covariance matrix. They also developed an alternating optimization algorithm that iteratively optimizes one of the reflector coefficients or the transmit covariance matrix while fixing other variables, deriving the optimal solution for each subproblem and significantly reducing computational complexity. However, the results show that the proposed algorithm can only guarantee convergence to a local optimum. Wang Lu et al., in their paper "Optimization for Maximizing System Throughput in IRS-NOMA Systems" published in *Computer Engineering and Design*, significantly enhanced system throughput by employing a distributed deployment of intelligent reflectors based on reflector phase discretization. However, this came at the cost of increased system complexity and the need to manage more reflector units, leading to increased energy consumption and cost.

[0004] Existing literature indicates that common methods to improve the performance of intelligent reflector communication systems involve optimizing the reflection characteristics of reflector units and changing the deployment method of reflectors. However, these studies have neglected the sparsity design of intelligent reflectors. By carefully selecting the positions of reflector units, sparsified intelligent reflectors can optimize signal paths, thereby improving signal quality and increasing system capacity. Therefore, based on existing technologies, the performance of intelligent reflectors can be further optimized. Furthermore, the selection of reflector units can be combined with current swarm intelligence optimization algorithms. Through algorithm iteration, the optimal solution to the problem is searched within the solution space, greatly simplifying computational complexity and improving problem-solving efficiency. Currently, the swarm intelligence optimization algorithms used for discrete optimization problems are mostly early algorithms such as particle swarm optimization and genetic algorithms. These algorithms have relatively simple search mechanisms and suffer from slow convergence speed and low convergence accuracy. In his paper "Hunter-prey optimization: algorithm and applications" published in *Soft Computing* (2022, Vol. 26, pp. 1279-1314), Iraj Naruei proposed the deer-hunting algorithm, which consists of two phases: exploration and development. In the exploration phase, the prey is driven to move towards unexplored areas by the hunter's predatory behavior, ensuring the algorithm's global search capability. In the development phase, the prey moves towards the safest location—the region with the highest fitness value—to avoid the hunter, thus guiding the algorithm to converge quickly. Therefore, this algorithm has good global search capability and can converge to a near-optimal solution relatively quickly. However, the traditional deer-hunting algorithm can only solve continuous engineering problems and cannot be directly applied to the sparse problem of intelligent reflective surfaces. This invention designs a quantum deer-hunting algorithm based on the original algorithm, overcoming the limitations of its application. The single-chain encoded quantum deer hunting method updates the quantum rotation angle and quantum velocity based on the original position of each deer and the current local and global optimal solutions. Then, it measures the original position, transforming the deer hunting position from a continuous vector into a discrete vector. This ensures the randomness of the position update in each generation and preserves the global search capability of the deer hunting algorithm. Summary of the Invention

[0005] The purpose of this invention is to provide a quantum deer-hunting mechanism-based intelligent reflector sparsity method. This method effectively reduces the number of reflector elements while maintaining the performance of the intelligent reflector. It automatically optimizes the layout of reflector elements using a swarm intelligence optimization algorithm to find the optimal solution. Furthermore, this invention employs a more advanced deer-hunting algorithm. Addressing the limitation that this algorithm can only solve continuous engineering problems and cannot be directly applied to intelligent reflector sparsity solutions, this invention combines quantum encoding with the original algorithm to discretize it, proposing a quantum deer-hunting method. The designed intelligent reflector sparsity scheme based on the quantum deer-hunting mechanism possesses better global search capabilities and higher convergence accuracy.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] A method for sparse intelligent reflective surfaces in a quantum deer-hunting mechanism, the specific steps of which are as follows:

[0008] Step 1: Model the signal transmission system, which includes a base station, a smart reflector, and a receiver, to obtain the signal received by the receiver;

[0009] Step 2: Model the system capacity calculation method for the sparse smart reflector to obtain the information transmission rate of the receiver after sparseness.

[0010] Step 3: Initialize the quantum velocity and position of each deer in the deer population, with each deer's position consisting of "0" and "1"; and iterate until the optimal position is reached;

[0011] Step 4: For the nth deer, generate a random number uniformly distributed between 0 and 1. When the probability of the hunting deer is less than the switching probability β, the nth hunting deer performs the exploration behavior; otherwise, it performs the development behavior.

[0012] Step 5: Perform the exploration behavior; after discretizing the deer hunting locations, determine whether the number of "1"s in the deer hunting locations exceeds L. If it does, then...

[0013] Step Six: Execute the development behavior; after discretizing the hunting deer positions, determine whether the number of "1"s in the hunting deer exceeds L. If it does, then...

[0014] Step 7: Record the position of each deer after the update, input the fitness function to calculate the fitness value, and update the global optimum and local optimum positions;

[0015] Step 8: Determine if the maximum number of iterations has been reached. If not, let t = t + 1 and return to step 4 to continue iterating; otherwise, output the global optimal position to obtain the sparse scheme of intelligent reflective surface reflective element based on quantum deer hunting mechanism.

[0016] Furthermore, step one specifically includes:

[0017] Suppose the base station has A antennas, the receiver has B antennas, and the smart reflector has a maximum of D reflecting elements; the base station's transmission power matrix can be represented as:

[0018]

[0019] Where diag{·} denotes a diagonal matrix, p a This represents the transmission power of the a-th antenna at the base station, where a = 1, 2, ..., A, and 0 ≤ p. a ≤p max p max This indicates the maximum power that each base station antenna is allowed to transmit;

[0020] The matrix formed by the reflection phases of each reflection unit in the intelligent reflective surface is represented as: Let d represent the reflection phase of the d-th reflective element, where d = 1, 2, ..., D, and then obtain the reflection phase matrix of the intelligent reflective surface reflective unit. in The reflection coefficient;

[0021] The channel state information from the base station to the receiver is represented as: The channel state information from the base station to the smart reflector is represented as follows: The channel state information from the smart reflector to the receiver is represented as follows: This represents the set of all channel state information matrices of size A×B;

[0022] The unit energy signal transmitted by the base station is used The complex Gaussian white noise of the smart reflector and the receiver is expressed as follows: and Then the signal received by the receiving end is obtained:

[0023] g REC =(F BS-REC +F IR-REC ·ψ·F BS-IR )·P·e+F IR-REC ·ψ·n IR +n REC

[0024] in,

[0025] Furthermore, step two specifically involves:

[0026] The sparse vector representation of the smart reflective surface is as follows: X = [x1, x2, ..., x D Each element in the array can only be either "0" or "1". "1" represents setting a reflective element at that location, while "0" represents not setting a reflective element at that location.

[0027] Suppose that the smart reflective surface is sparsed with sparsity δ. Then the smart reflective surface should have L = round(δD) reflective units, where round means rounding to the nearest integer, that is, X can only have L "1"s.

[0028] The information transmission rate at the receiving end after sparsening is:

[0029]

[0030] Here, det(·) represents the value of the determinant of a matrix, (·) H The symbol represents the conjugate transpose of a matrix; ⊙ represents the multiplication of corresponding elements of the two vectors before and after the symbol; Ι B The total noise matrix is ​​a B×B dimensional identity matrix. REC Calculated by the following formula:

[0031]

[0032] Where k1 represents the noise power at the smart reflector and k2 represents the noise power at the receiver.

[0033] The system capacity of a large-scale multiple-input multiple-output communication system is expressed as: That is, the information transmission rate at the receiving end.

[0034] Furthermore, step three specifically includes:

[0035] System capacity As the fitness function of the quantum deer hunting algorithm, the fitness value is calculated, and the sparse matrix X of the intelligent reflective surface is used as the optimization vector, that is, the position of the deer hunted.

[0036] Let N be the number of deer in the deer population, T be the maximum number of iterations, and D be the search dimension. Let the quantum speed of the nth deer in the t-th iteration be . The position of the nth deer is: Where t is the number of iterations;

[0037] Substituting the position of the nth hunting deer into the fitness function, the higher the fitness value, the better the hunting deer's position. The hunting deer position with the highest fitness value found up to generation t is taken as the globally optimal position in generation t. The position with the highest fitness value found up to generation t for the nth deer is taken as the local optimum position for the nth deer in generation t.

[0038] According to the formula: The measurement rules were applied to the first-generation quantum velocity measurement to obtain the discrete positions of each deer in the first-generation population, where... Let d be a random number generated for the nth hunting deer in the first generation, where d = 1, 2, ..., D; calculate the fitness value of each hunting deer in the first generation, and take the position of the hunting deer with the highest fitness value in the first generation as the global optimal position in the first generation. The position of the nth hunting deer in the first generation is taken as its own local optimum position in the first generation.

[0039] Furthermore, the exploration behavior in step four refers to behavior that tends to be highly random. In this behavior, the hunter's position is updated to encourage the prey to move away from the hunter and explore unexplored areas. The development behavior, after discovering a promising area, must reduce random behavior and be able to search around the promising area. In this behavior, the prey's position is updated to encourage the prey to move towards the globally optimal position.

[0040] Furthermore, step five specifically includes:

[0041] The hunter's position is updated according to the following formula:

[0042]

[0043] Among them, C t It is a balance parameter, derived from Updated, its value decreases from W1 to W2 during the iteration process; μ t It is the average position of all prey in generation t, i.e. It is a coordination parameter, the value of which is determined by... The formula is determined, where and It is an N×1 dimensional vector, and its elements are uniformly randomly distributed between [0,1]. It is an N×1 dimensional column vector, by The measurement rules are obtained, where Representing vectors The d-th element, It is a variable that follows a uniform random distribution between [0, 1], representing The d-th element; It is the hunter's prey in generation t, in order to determine First, calculate the distance from the nth deer to μ. t Euclidean distance: in μ t The d-th element; then... Sort the elements in the array from smallest to largest, and finally let round indicates rounding down to the nearest integer. The sorted number The element represents the deer as the prey of the t-th generation hunter;

[0044] The quantum rotation angle of the d-th dimension of the nth hunting deer in the (t+1)th generation is The calculation process for the quantum velocity and measurement position of the nth hunting deer in the (t+1)th generation in the d-th dimension is as follows:

[0045]

[0046] in, Let e1 represent the random number generated in the d-th dimension of the nth hunting deer in the (t+1)th generation; e1 and e2 are the local learning factor weight and the global learning factor weight, respectively, and both are constants. The mutation probability, The random numbers are uniformly distributed between [0, 1]. This represents the locally optimal position found by the nth hunting deer up to the tth generation. This represents the globally optimal position found by the entire population up to generation t, where abs indicates taking the absolute value.

[0047] After discretizing the deer hunting locations, determine if the number of "1"s in each location exceeds L. If it does, then...

[0048] Furthermore, step six specifically includes:

[0049] According to the formula Update prey location and balance parameters The update formula for the adjustment parameter Z is the same as that in the exploration phase; It is a random number that is uniformly and randomly distributed in the range [0, 1].

[0050] The d-dimensional quantum rotation angle of the nth hunting deer in generation t+1 is... The calculation process for the d-th dimension quantum velocity and measurement position of the nth hunting deer in the (t+1)th generation is as follows:

[0051]

[0052] in, Let e1 represent the random number generated in dimension d by the nth hunting deer in generation t+1; e1 and e2 are the local weight and global weight, respectively, and both are constants. The mutation probability, The random numbers are uniformly distributed between [0, 1]. This represents the locally optimal position found by the nth hunting deer up to the tth generation. Let L be the globally optimal position found by the entire population up to generation t; abs represents taking the absolute value; after discretizing the hunting positions, determine whether the number of "1"s in the hunting positions exceeds L. If it does, then...

[0053] Furthermore, step seven specifically includes:

[0054] The updated deer locations are then fed into the fitness function to calculate the fitness value for each deer. if but Keep it unchanged; otherwise, let

[0055] Record the fitness value of each deer in generation t+1. Sort all positions searched from the nth deer up to the t+1th generation according to their fitness values ​​in descending order, and take the position of the deer with the highest fitness value as the local optimum position of the nth deer. Sort all deer hunting locations found in generation t+1 by fitness value from largest to smallest, and select the deer hunting location with the largest fitness value as the globally optimal location in generation t+1.

[0056] A computer device / apparatus / system includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of a smart reflective surface sparsity method for a quantum deer hunting mechanism.

[0057] A computer-readable storage medium having a computer program / instructions thereon, which, when executed by a processor, implements steps of a smart reflective surface sparse method for a quantum deer hunting mechanism.

[0058] The beneficial effects of this invention are as follows:

[0059] (1) Traditional deer hunting algorithms can only solve continuous optimization problems and cannot be directly applied to the sparse problem of intelligent reflective surfaces. To address this issue, this invention combines the traditional deer hunting algorithm with quantum coding to design a quantum deer hunting mechanism, thus overcoming the application limitations of the original algorithm.

[0060] The single-chain encoded quantum deer-hunting method updates the quantum rotation angle and quantum velocity based on the original position, local optimal position, and global optimal position of each deer. The velocity is then measured to obtain a discretized position, which can be applied to the sparsity problem of intelligent reflective surfaces. Simultaneously, it ensures the randomness of position updates in each generation, preserving the global search capability of the deer-hunting mechanism.

[0061] (2) Previous studies on improving the performance of intelligent reflective surface communication systems have neglected the sparse design of intelligent reflective surfaces. In addition to improving the reflection characteristics of reflective units and the layout of intelligent reflective surfaces, this invention provides an intelligent reflective surface sparse scheme that can effectively improve the performance of intelligent reflective surfaces. The scheme automatically optimizes the layout of reflective units through the quantum deer hunting mechanism to find the optimal layout of reflective elements.

[0062] After searching using the quantum deer-hunting mechanism, the performance of the intelligent reflective surface was significantly improved. Compared with some classic swarm intelligence optimization methods such as particle swarm optimization, the sparse scheme of intelligent reflective surfaces based on the quantum deer-hunting mechanism shows superior search accuracy, demonstrating the strong practical value of this invention in the field of sparse intelligent reflective surfaces. Attached Figure Description

[0063] Figure 1 Flowchart of the optimal sparse method for intelligent reflective surfaces based on quantum deer hunting mechanism;

[0064] Figure 2 Performance comparison curves of quantum deer hunting algorithm and particle swarm optimization algorithm under 50% sparsity;

[0065] Figure 3 Performance comparison curves of quantum deer hunting algorithm and particle swarm optimization algorithm under 80% sparsity;

[0066] Figure 4 A graph comparing system capacity under 50% sparsity and non-sparse conditions. Detailed Implementation

[0067] The present invention will now be further described with reference to the accompanying drawings.

[0068] according to Figure 1 The present invention discloses a smart reflective surface sparse method for a quantum deer hunting mechanism, the specific steps of which are as follows:

[0069] Step 1: Model the signal transmission system, which includes the base station, smart reflector, and receiver.

[0070] Assume the base station has A antennas, the receiver has B antennas, and the smart reflector has a maximum of D reflecting elements. The base station's transmission power can be represented as a matrix. Where diag{·} denotes a diagonal matrix, p a This represents the transmission power of the a-th antenna at the base station, where a = 1, 2, ..., A, and 0 ≤ p. a ≤p max p max This represents the maximum transmit power allowed for each base station antenna. The matrix formed by the reflection phases of each reflecting element in the smart reflector is represented as... Let d represent the reflection phase of the d-th reflective element, where d = 1, 2, ..., D, and then obtain the reflection phase matrix of the intelligent reflective surface reflective unit. in The reflection coefficient is used. The channel state information from the base station to the receiver is represented as... The channel state information from the base station to the smart reflector is represented as follows: The channel state information from the smart reflector to the receiver is represented as follows: This represents the set of all channel state information matrices of size A×B.

[0071] The unit energy signal transmitted by the base station is used The complex Gaussian white noise of the smart reflector and the receiver is expressed as follows: and Then the signal received by the receiver can be obtained:

[0072] g REC =(F BS-REC +F IR-REC ·ψ·F BS-IR )·P·e+F IR-REC ·ψ·n IR +n REC

[0073] in,

[0074] Step 2: Model the system capacity calculation method for sparse intelligent reflective surfaces.

[0075] The sparse vector of the smart reflective surface is represented as X = [x1, x2, ..., x D Each element in X can only be either "0" or "1", where "1" represents a reflective element at that location, and "0" represents no reflective element at that location. Assuming a sparse scheme with a sparsity of δ is applied to the smart reflective surface, the smart reflective surface should then have L = round(δD) reflective units, where round represents rounding down to the nearest integer, meaning that X can only contain L "1"s. The information transmission rate at the receiving end after sparsening is:

[0076]

[0077] Here, det(·) represents the value of the determinant of a matrix, (·) H The symbol represents the conjugate transpose of a matrix; ⊙ represents the multiplication of corresponding elements of the two vectors before and after the symbol; Ι B The total noise matrix is ​​a B×B dimensional identity matrix. REC Calculated by the following formula:

[0078]

[0079] Where k1 represents the noise power at the smart reflector and k2 represents the noise power at the receiver. The system capacity of a large-scale multiple-input multiple-output communication system is expressed as: That is, the information transmission rate at the receiving end.

[0080] Step 3: Initialize the quantum velocity and position of each deer in the hunting population. Each deer's position consists of "0"s and "1s". Set the system capacity... As the fitness function of the quantum deer hunting algorithm, the fitness value is calculated, and the sparse matrix X of the intelligent reflective surface is used as the optimization vector, that is, the position of the deer hunt.

[0081] Let N be the number of deer in the deer population, T be the maximum number of iterations, and D be the search dimension. Let the quantum speed of the nth deer in the t-th iteration be . The position of the nth deer is Where t is the iteration number. Substituting the position of the nth hunting deer into the fitness function, a higher fitness value indicates a better hunting deer position. The hunting deer position with the highest fitness value found up to generation t is taken as the globally optimal position in generation t. The position with the highest fitness value found up to generation t for the nth deer is taken as the local optimum position for the nth deer in generation t. according to The measurement rules were applied to the first-generation quantum velocity measurement to obtain the discrete positions of each deer in the first-generation population, where... Let d represent a random number generated for the nth hunting deer in the first generation, where d = 1, 2, ..., D. Calculate the fitness value of each hunting deer in the first generation, and take the position of the hunting deer with the highest fitness value in the first generation as the globally optimal position in the first generation. The position of the nth hunting deer in the first generation is taken as its own local optimum position in the first generation.

[0082] Step 4: For the nth deer, generate a random number uniformly distributed between 0 and 1. When the probability is less than the switching probability β, the nth hunting deer performs the exploration behavior; otherwise, it performs the development behavior.

[0083] Exploration behavior refers to actions that tend towards high randomness, thus leading to significant variations in the solution. In this behavior, the hunter's position is updated, prompting the prey to move away from the hunter and explore unexplored areas. Once a promising area is discovered, random behavior must be reduced to allow for searching around that area; this is development. In this behavior, the prey's position is updated, prompting the prey to move towards the globally optimal location.

[0084] Step 5: Perform the exploration action. Update the hunter's position according to the formula, as follows:

[0085]

[0086] Among them, C t It is a balance parameter, derived from The value is updated, decreasing from W1 to W2 during the iteration. μ t It is the average position of all prey in generation t, i.e. It is a coordination parameter, the value of which is determined by... The formula is determined, where and It is an N×1 dimensional vector, and its elements are uniformly randomly distributed between [0,1]. It is an N×1 dimensional column vector, by The measurement rules are obtained, where Representing vectors The d-th element, It is a variable that follows a uniform random distribution between [0,1], representing The d-th element. It is the hunter's prey in generation t, in order to determine First, calculate the distance from the nth deer to μ. t Euclidean distance: in μ t The d-th element. Then... Sort the elements in the array from smallest to largest, and finally let round indicates rounding down to the nearest integer. The sorted number The element represents the deer as the prey of the t-th generation hunter.

[0087] The quantum rotation angle of the d-th dimension of the nth hunting deer in the (t+1)th generation is The calculation process for the quantum velocity and measurement position of the nth hunting deer in the (t+1)th generation in the d-th dimension is as follows: in Let represent the random number generated in the d-th dimension by the nth hunting deer in the (t+1)-th generation. e1 and e2 are the weights of the local learning factor and the global learning factor, respectively, and both are constants. The mutation probability, is a random number that is uniformly distributed between [0, 1]. This represents the locally optimal position found by the nth hunting deer up to the tth generation. Let abs represent the globally optimal position found by the entire population up to generation t, where abs indicates taking the absolute value. After discretizing the hunting positions, determine if the number of "1"s in the hunting positions exceeds L. If it does, then...

[0088] Step Six: Perform development actions. According to the formula... Update prey location and balance parameters The update formula for the adjustment parameter Z is the same as that for the exploration phase. It is a random number that is uniformly and randomly distributed in the range [0, 1].

[0089] The d-dimensional quantum rotation angle of the nth hunting deer in generation t+1 is... The calculation process for the d-th dimension quantum velocity and measurement position of the nth hunting deer in the (t+1)th generation is as follows: in Let e1 represent the random number generated in dimension d of the nth hunting deer in the (t+1)th generation. e1 and e2 are the local weight and global weight, respectively, and both are constants. The mutation probability, is a random number that is uniformly distributed between [0, 1]. This represents the locally optimal position found by the nth hunting deer up to the tth generation. Let L be the globally optimal position found by the entire population up to generation t. `abs` represents taking the absolute value. After discretizing the hunting positions, determine if the number of "1"s in the hunting positions exceeds L. If it does, then...

[0090] Step 7: Record the position of each deer after the update, input the fitness function to calculate the fitness value, and update the global optimal position and the local optimal position.

[0091] The updated deer locations are then fed into the fitness function to calculate the fitness value for each deer. if but Keep it unchanged; otherwise, let Record the fitness value of each deer in generation t+1. Sort all positions searched from the nth deer up to the t+1th generation according to their fitness values ​​in descending order, and take the position of the deer with the highest fitness value as the local optimum position of the nth deer. Sort all deer hunting locations found in generation t+1 by fitness value from largest to smallest, and select the deer hunting location with the largest fitness value as the globally optimal location in generation t+1.

[0092] Step 8: Determine if the maximum number of iterations has been reached. If not, let t = t + 1 and return to step 4 to continue iterating; otherwise, output the global optimal position and obtain the sparse scheme of intelligent reflective surface reflective element based on the quantum deer hunting mechanism.

[0093] according to Figures 2 to 3 To comprehensively compare the performance of the two methods, this invention performs the same initialization on both methods, with a noise figure k1 = 10. -11k2 = 10 -11 Reflectance coefficient The parameters of the basic quantum deer hunting algorithm are set to β = 0.1. e1 = 0.2, e2 = 0.8, α1 = 0.5, α2 = 0.98, w1 = 1, w2 = 0.02. The Particle Swarm Optimization (PSO) algorithm was proposed by J. Kennedy et al. in their paper "Particle swarm optimization" published in *Proceedings of ICNN'95 - International Conference on Neural Networks*. The parameters of the PSO algorithm are c1 = 0.2. e1 = 0.2, e2 = 0.8. The population size and maximum number of iterations for both search mechanisms are the same: N = 50, T = 500, and the number of reflective elements is L = 40. At 50% and 80% sparsity, the hunting dimensions for both search mechanisms are D = 80 and D = 50, respectively. The presence or absence of a reflective element in each hunted deer and each particle is represented by 1 and 0. Each method is run independently 150 times, and the average fitness from these 150 runs is used to plot a fitness curve. The hunting position with the optimal fitness is taken as the best sparse scheme for the intelligent reflective surface.

[0094] Simulation results of fitness curves for the same number of reflective elements at 50% sparsity are as follows: Figure 2 As shown, the average value of 150 runs is plotted. The simulation results of the fitness curves for the same number of reflective elements at 80% sparsity are as follows. Figure 3 As shown, the average value of 150 runs is plotted. It can be seen that, under the same sparsity, the quantum deer hunting mechanism (HPO) has the advantage of higher convergence accuracy compared to the particle swarm optimization (PSO) algorithm, fully demonstrating that the quantum-encoded deer hunting mechanism can effectively solve the discrete optimization problem of sparsity in intelligent reflectors. According to... Figure 4 It can be seen that the performance of the smart reflective surface is significantly improved after sparsening compared to the non-sparse state.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for sparse intelligent reflective surfaces in a quantum deer hunting mechanism, characterized in that: The specific steps are as follows: Step 1: Model the signal transmission system, which includes a base station, a smart reflector, and a receiver, to obtain the signal received by the receiver; Step 2: Model the system capacity calculation method for the sparse smart reflector to obtain the information transmission rate of the receiver after sparseness. Step 3: Initialize the quantum velocity and position of each deer in the deer population, with each deer's position consisting of "0" and "1"; and iterate to the optimal position; Step 4: For the first Hunt only deer and generate random numbers uniformly distributed between 0 and 1. When it is less than the switching probability Time They only hunt deer to carry out exploratory behavior; Otherwise, proceed with the development process; Exploration behavior refers to behavior that tends to be highly random. In this behavior, the hunter's position is updated, causing the prey to move away from the hunter and explore unexplored areas. Development behavior, on the other hand, is behavior that, after discovering a promising area, must reduce random behavior and be able to search around the promising area. In this behavior, the prey's position is updated, causing the prey to move towards the globally optimal position. Step 5: Perform the exploration behavior; after discretizing the deer hunting locations, determine whether the number of "1"s in the deer hunting locations exceeds [a certain threshold]. If it exceeds, then ; The hunter's position is updated according to the following formula: ； in, It is a balance parameter, derived from Update, its value changes during the iteration process from Reduce to ; It is the first The average position of all prey in the generation, i.e. ; It is a coordination parameter, the value of which is determined by... The formula is determined, where and yes A vector of dimension, wherein each element follows a uniform random distribution between [0,1]. yes A column vector of dimension, by The measurement rules are obtained, where Representing vectors The Middle One element, It is a variable that follows a uniform random distribution between [0, 1], representing The Middle One element; It is the first The hunter's prey, in order to determine First calculate the first Only hunting deer Euclidean distance: ,in express The Middle Dimensional elements; then... Sort the elements in the array from smallest to largest, and finally let , This indicates rounding down to the nearest integer. The sorted number The element representing the deer hunting as the first The hunter's prey; No. The generation The first deer hunt The quantum rotation angle of dimension is , No. The generation The first deer hunt The calculation process for the quantum velocity and measurement position of the dimension is as follows: , ； in, Indicates the first The middle generation The first deer hunt Dimensionally generated random numbers; and These are the local learning factor weights and the global learning factor weights, respectively, and both are constants; The mutation probability, To distribute evenly in Random numbers between; For the first up to the generation Only the locally optimal location found by the deer hunting search. For the first The globally optimal position found by the entire population up to generation. Indicates taking the absolute value; After discretizing the deer hunting locations, determine whether the number of "1"s in the deer hunting locations exceeds [a certain threshold]. If it exceeds, then ; Step Six: Execute development actions; after discretizing the hunting locations, determine whether the number of "1"s in the hunting locations exceeds [a certain threshold]. If it exceeds, then Wherein, the sparsity of the intelligent reflective surface is assumed to be... In the sparse scheme, the smart reflective surface should then have One reflective unit, This means rounding down to the nearest integer. Only one is allowed in the middle. One "1"; For the first During the nth iteration Location of only deer hunting; according to the formula Update prey location. yes A random number that is uniformly and randomly distributed within a range; Step 7: Record the position of each deer after the update, input the fitness function to calculate the fitness value, and update the global optimum and local optimum positions; Step 8: Determine if the maximum number of iterations has been reached. If not, let... If the iteration continues, return to step four; otherwise, output the global optimal position to obtain a sparse scheme for intelligent reflective surface reflective elements based on the quantum deer hunting mechanism.

2. The intelligent reflective surface sparse method for a quantum deer hunting mechanism according to claim 1, characterized in that: Step one specifically involves: Assume the base station has One antenna, the receiver has The antenna has a maximum of [number] smart reflectors. One reflection unit; the matrix representation of the base station transmission power is as follows: ； in, Represents a diagonal matrix. That is, the base station number The transmission power of the root antenna, ,and , This indicates the maximum power that each base station antenna is allowed to transmit; The matrix formed by the reflection phases of each reflection unit in the intelligent reflective surface is represented as: , Indicates the first The reflection phase of each reflecting element, This leads to the reflection phase matrix of the intelligent reflective surface reflection unit. ,in The reflection coefficient; The channel state information from the base station to the receiver is represented as: The channel state information from the base station to the smart reflector is represented as follows: The channel state information from the intelligent reflector to the receiver is represented as follows: ; The unit energy signal transmitted by the base station is used The complex Gaussian white noise of the smart reflector and the receiver is expressed as follows: and Then the signal received by the receiving end is obtained: ； in, .

3. The intelligent reflective surface sparse method for a quantum deer hunting mechanism according to claim 1, characterized in that: Step two specifically involves: The sparse vector representation of the smart reflective surface is as follows: , Each element can only take the value "0" or "1". "1" means that a reflective element is set at that location, and "0" means that no reflective element is set at that location. Let the sparsity of the smart reflective surface be... In the sparse scheme, the smart reflective surface should then have One reflective unit, This means rounding down to the nearest integer. Only one is allowed in the middle. One "1"; The information transmission rate at the receiving end after sparsening is: ； in, This indicates finding the value of the determinant of a matrix. Represents the conjugate transpose of a matrix. This indicates the multiplication of corresponding elements in the two vectors before and after the symbol. for 3D identity matrix, total noise matrix Calculated by the following formula: ； in, This indicates the noise power at the smart reflector. This indicates the noise power at the receiving end; The system capacity of a large-scale multiple-input multiple-output communication system is expressed as: That is, the information transmission rate at the receiving end.

4. The intelligent reflective surface sparse method for a quantum deer hunting mechanism according to claim 3, characterized in that: Step three specifically involves: System capacity As the fitness function of the quantum deer hunting algorithm, the fitness value is calculated, and the sparse matrix of the intelligent reflective surface is used. As the optimization vector, that is, the position of the deer hunt; Suppose that the number of hunting deer in the hunting deer population is . The maximum number of iterations is , Let the search dimension be the first... During the nth iteration The quantum speed of hunting deer is , , , , No. The location for hunting deer is: ,in, This represents the number of iterations. The first Substituting the hunting location into the fitness function, the higher the fitness value, the better the hunting location; this continues until the... The position with the highest fitness value found so far is used as the first deer hunting position. Replace the global optimal position ; will the first Hunt deer until the 1st The position with the highest fitness value found so far is taken as the first position. Hunting deer Local optimal position of the generation ; According to the formula: The measurement rules were applied to the first-generation quantum velocity measurement to obtain the discrete positions of each deer in the first-generation population, where... Indicates the first The middle generation The first deer hunt Generated random numbers in dimension Calculate the fitness value of each hunting deer in the first generation, and take the position of the hunting deer with the highest fitness value in the first generation as the global optimal position in the first generation. The first generation The location of the deer hunt is taken as its first-generation local optimum. .

5. The intelligent reflective surface sparse method for a quantum deer hunting mechanism according to claim 1, characterized in that: Step seven specifically involves: The updated deer locations are then fed into the fitness function to calculate the fitness value for each deer. ;if ,but Keep it unchanged; otherwise, let , ; Record number The fitness value of each deer in the generation, for the first Hunting deer to the first Sort all locations found so far by fitness value from largest to smallest, and select the deer hunting location with the largest fitness value as the [number]th [position]. Local optimal position for hunting deer ; for the first All deer hunting locations found through the search are sorted in descending order of fitness value, and the deer hunting location with the highest fitness value is selected as the [number]th [position]. Replace the global optimal position .

6. A computer system comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that: When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 5.

Images