System and method for detecting clustered active users based on particle swarm optimization algorithm

Abstract

The invention discloses a clustered active user detection system and method based on a particle swarm optimization algorithm, and belongs to the technical field of active user detection methods. The method comprises the following steps: S1, designing an objective function, and evaluating cluster division quality by using the objective function; different weight coefficients are introduced to adjust an objective function so as to adapt to different scene requirements; s2, introducing a large-scale fading coefficient variance constraint into a system optimization process to ensure that the quality of a channel between an access point in a cluster and a user is uniform; s3, based on a particle swarm optimization function module, dynamically optimizing cluster number selection through a particle swarm optimization algorithm, and further realizing cluster number selection optimization through a fitness function calculation module and a large-scale fading coefficient variance calculation module; and S4, after optimization is completed, outputting an optimal cluster number and a corresponding fitness value through a particle swarm optimization algorithm. Compared with the prior art, the active user detection method is suitable for active user detection in a cellular-removed large-scale multiple-input-multiple-output network, and the detection precision and the calculation efficiency are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of active user detection methods, and particularly to a clustering active user detection system and method based on a particle swarm optimization algorithm. Background Art

[0002] The adaptive cluster number selection scheme has significant advantages in dynamic network environments such as mobile networks, wireless sensor networks, and massive machine-type communications due to its dynamic adjustment characteristics. This scheme can adjust the cluster number in real time according to factors such as network load, user distribution, and communication requirements, thereby effectively improving resource utilization efficiency. In a cellular-free massive multiple-input multiple-output (Massive MIMO) system, the number of devices is large and the distribution is intricate, and traditional static cluster number selection methods are difficult to adapt to this complex and changeable environment. The adaptive cluster number selection scheme can optimize the cluster number according to the real-time situation and requirements of the network, significantly enhancing the scalability and reliability of the system.

[0003] In the field of communication, adaptive cluster number selection algorithms include genetic algorithms, ant colony optimization algorithms, simulated annealing algorithms, and reinforcement learning algorithms. For the adaptive cluster number selection of the K-means algorithm, the current mainstream schemes include K-means++ that improves the cluster center selection, and the adaptive selection method based on the silhouette coefficient. K-means++ effectively improves the clustering effect and reduces the probability of the algorithm converging to a local optimal solution by improving the initial cluster center selection strategy of the traditional K-means algorithm. The silhouette coefficient, as a key indicator for evaluating the quality of clustering results, ranges from -1 to 1. It comprehensively considers the intra-cluster compactness and inter-cluster separation. Specifically, the closer the silhouette coefficient is to 1, the more reasonable the distribution of data points within the cluster, with high intra-cluster similarity and large inter-cluster differences; if the silhouette coefficient is close to -1, it means that the data points may be misallocated and should be attributed to other clusters. The cluster number selection method based on the silhouette coefficient calculates the silhouette coefficient under different cluster numbers and takes the cluster number that maximizes the silhouette coefficient as the final selection, fully considering both intra-cluster compactness and inter-cluster separation.

[0004] However, in the scenarios of cellular-free massive MIMO and massive machine-type communications, there are still many problems with the optimization algorithms for cluster number adaptive selection and active user detection in clustering algorithms:

[0005] 1. Significant limitations in static settings: Most existing cluster number selection methods rely on static cluster numbers set artificially in advance. This approach has significant limitations in actual communication applications, especially in dynamic network environments. Due to the frequent changes in the number of devices and user loads in the network, fixed cluster number partitioning and location - based partitioning methods are difficult to adapt to these dynamic fluctuations. Too many clusters will cause resource waste, while too few will lead to resource shortages. In ultra - dense deployment scenarios, it may cause access point resource overload or some clusters to be idle, seriously affecting communication performance.

[0006] 2. Insufficient consideration of channel factors: In the scenarios of cell - free massive MIMO networks and massive machine - type communication, the channel quality difference between access points and users has a huge impact on system performance. Traditional cluster partitioning methods such as K - means fail to effectively consider large - scale fading coefficients and channel characteristics, resulting in uneven signal quality within clusters, which in turn affects the collaborative detection performance.

[0007] 3. Lack of dynamic response ability: Traditional cluster number selection algorithms cannot respond to network changes in real - time. In a dynamic environment, network load, signal quality, and user activities are constantly changing, while static cluster number selection methods cannot adjust the cluster number in a timely manner according to these changes. Therefore, when network traffic fluctuates violently, the cluster number selection often fails to reach the optimal, resulting in low system efficiency and poor active user detection performance.

[0008] 4. Excessive computational complexity: As the network scale expands and the number of devices increases in cell - free massive MIMO networks and massive machine - type communication scenarios, the computational complexity of cluster number selection and active user detection grows exponentially. Traditional cluster number selection methods, when dealing with large - scale networks, generate a large amount of computational overhead due to the frequent calculation of the distance between users and cluster centers, resulting in slow system response and poor real - time performance. During the active user detection process, complex cluster partitioning processes introduce unnecessary delays, reducing detection accuracy and efficiency.

[0009] To solve the above problems, the present invention proposes a clustering active user detection system and method based on the particle swarm optimization algorithm. Summary of the Invention

[0010] The object of the present invention is to provide a clustering active user detection system and method based on the particle swarm optimization algorithm to solve the problems mentioned in the background technology.

[0011] To achieve the above - mentioned object of the invention, the present invention provides the following technical solutions:

[0012] A clustering active user detection system based on the particle swarm optimization algorithm, comprising the following modules:

[0013] K-means Network Function Module: Using the K-means algorithm to cluster the geographical locations of access points and user devices, calculate the distance between users and access points, and assign access points to different clusters according to the distance; this module not only reduces the computational complexity but also provides input data for subsequent large-scale fading coefficient calculation and active user detection;

[0014] Covariance Matrix Optimization and Active Mode Decoding Function Module: Used to decode the active user mode and calculate the active parameters of users based on the received signals; by minimizing the negative log-likelihood function and optimizing the covariance matrix, gradually adjust the active state parameters of each user to accurately determine whether the user is active;

[0015] Large-scale Fading Coefficient Calculation Module: Calculate the large-scale fading coefficient within each cluster according to the distance between users and access points, and obtain the overall large-scale fading coefficient of each cluster through the weighted average method; this process helps the model more accurately evaluate the signal strength of different users and optimize subsequent active user detection;

[0016] Signature Sequence Generation Module: Generate a unique signature sequence for identifying each user. By generating a signature sequence with a complex Gaussian distribution, provide a unique signal identifier for user devices to ensure the accuracy of signal decoding and active state detection;

[0017] Complex Gaussian Noise Generation Module: Used to simulate the noise in the wireless channel, generate complex Gaussian noise and add it to the received signals; in this way, the system can simulate the actual wireless communication environment and help evaluate the active user detection performance under different signal-to-noise ratio conditions;

[0018] Miss Detection and False Alarm Probability Calculation Module: Used to calculate two key performance indicators of the active user detection system, namely the miss detection probability and the false alarm probability, and adjust the detection results according to different thresholds; through this module, the system can evaluate the detection performance under different threshold settings, thereby optimizing the accuracy of active user detection;

[0019] Constant Parameter Calculation Module: Responsible for calculating polynomial coefficients for decoding the active state of users; by calculating polynomial coefficients related to the large-scale fading coefficient, optimize the decoding process of the system and provide mathematical support for the accurate estimation of the user's active state;

[0020] Particle Swarm Optimization Function Module: Based on the particle swarm optimization algorithm, implement dynamic optimization of the cluster number selection;

[0021] Fitness Function Calculation Module: Used to calculate the fitness value under a given number of clusters, and the fitness value comprehensively considers multiple factors such as the miss detection probability, the false alarm probability, the computational complexity, and the variance of the large-scale fading coefficient;

[0022] Large-scale fading coefficient variance calculation module: used to calculate the variance of the large-scale fading coefficient within each cluster to measure the uniformity of the signal quality within the cluster;

[0023] Main program module: responsible for initializing parameters and performing Monte Carlo simulation, setting the range of cluster number selection according to different numbers of access points, and running the particle swarm optimization algorithm for cluster number optimization.

[0024] Preferably, the specific execution process of the particle swarm optimization function module is as follows:

[0025] In particle swarm optimization, the position of the particle represents the number of clusters, and the velocity represents the change in the number of clusters; each particle adjusts its position and velocity according to the fitness value of the current number of clusters;

[0026] The particle updates its personal best and global best solutions during the search process and updates its velocity according to the inertia weight;

[0027] The particle swarm optimization algorithm obtains the optimal number of clusters through multiple iterations.

[0028] Preferably, the specific execution process of the fitness function calculation module is as follows:

[0029] For a given number of clusters, calculate the large-scale fading coefficient between the access points and users within the cluster;

[0030] Calculate the variance of the large-scale fading coefficient to reflect the uniformity of the signal quality within the cluster;

[0031] Calculate the probability of missed detection and false alarm;

[0032] Weight the above calculated indicators to obtain the fitness value of this number of clusters.

[0033] Preferably, the specific execution process of the large-scale fading coefficient variance calculation module is as follows:

[0034] Calculate the large-scale fading coefficient between the access points and users according to the cluster allocation, and calculate the weighted average value based on the large-scale fading coefficient;

[0035] Calculate the variance of the large-scale fading coefficient, reflecting the signal quality difference of the access points within each cluster: if the variance of the large-scale fading coefficient of a certain cluster is small, it indicates that the signal quality of the access points within the cluster is uniform, which is beneficial to cooperative detection; if the variance is large, it means that the signal quality difference of the access points within the cluster is large, and the detection effect is poor.

[0036] Preferably, the specific execution process of the main program running module is as follows:

[0037] For each number of access points, the main program optimizes the number of clusters by calling the particle swarm optimization function and calculates the large-scale fading coefficient and complexity performance indicators;

[0038] Through multiple experiments, cumulative calculation results are obtained to derive the average optimal number of clusters, computational complexity, and variance of large-scale fading coefficients under different APs and numbers of access points;

[0039] The main program generates charts to show the relationship between the number of clusters, complexity, and variance of large-scale fading coefficients varying with the number of access points, for analyzing and optimizing network design.

[0040] A method for detecting active users in clusters based on the particle swarm optimization algorithm implemented by using the system, comprising the following steps:

[0041] S1. Design an objective function that comprehensively considers the missed detection probability, false alarm probability, computational complexity, and variance of large-scale fading coefficients, and use it to evaluate the quality of cluster partitioning; different weight coefficients are introduced during the evaluation process to adjust the objective function to adapt to different scenario requirements;

[0042] S2. Introduce the variance constraint of large-scale fading coefficients into the system optimization process to ensure the uniform channel quality between access points and users within the cluster and improve the cooperative detection ability of the detection system;

[0043] S3. Based on the particle swarm optimization function module, dynamically optimize the selection of the number of clusters through the particle swarm optimization algorithm, and further realize the optimization of the selection of the number of clusters through the fitness function calculation module and the variance calculation module of large-scale fading coefficients;

[0044] S4. After the optimization is completed, output the optimal number of clusters and the corresponding fitness value through the particle swarm optimization algorithm, optimize the cluster partitioning in the active user detection work, and realize the adaptive selection of the number of clusters.

[0045] Preferably, the missed detection probability and false alarm probability affect the performance of the system for detecting active users; the computational complexity affects the resource consumption and latency of the system.

[0046] Preferably, the variance of large-scale fading coefficients is used to measure the balance of the channel quality between access points and users within the cluster: a low variance indicates uniform signal quality of access points within the cluster, which helps to improve the accuracy of cooperative detection; a high variance indicates large signal quality and affects the reliability of the system.

[0047] Preferably, the S3 specifically includes the following content:

[0048] Use the particle swarm optimization algorithm to simulate the process of a flock of birds foraging. Each particle represents a number of clusters, and continuously adjust the position and velocity of the particles through iteration to find the optimal number of clusters;

[0049] At each iteration, the particle updates its position and velocity according to the current fitness and the global optimal position to approach the global optimal solution. During the position update process of the particle, the constraint conditions of the objective function and the variance of the large-scale fading coefficient are considered, and a boundary processing mechanism is introduced to ensure that the particle always searches within the physically feasible range.

[0050] The fitness function is used to evaluate the quality of the cluster number selection, and the position of each particle is optimized according to the objective function and the constraint conditions.

[0051] By calculating the variance of the particle swarm fitness, it is judged whether the particle swarm has converged. When the fitness variance approaches zero, it indicates that the particle swarm has found the global optimal solution and the optimization process ends.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0053] (1) By dynamically selecting the cluster number based on the particle swarm optimization algorithm, the present invention can be adaptively adjusted according to the active user detection requirements, enabling the cluster number to flexibly respond to changes in network conditions. This dynamic adaptability significantly improves the network performance and scalability in the de-cellular massive multiple-input multiple-output network and the massive machine-type communication.

[0054] (2) The present invention introduces a multi-objective optimization function, comprehensively considering factors such as the missed detection probability, false alarm probability, computational complexity, and the variance of the large-scale fading coefficient into the optimization process. Through this method, the cluster number selection not only optimizes the geometric distribution of the clusters but also considers the communication quality such as the large-scale fading coefficient balance, thus achieving more efficient user detection and resource allocation.

[0055] (3) The present invention introduces a constraint on the variance of the large-scale fading coefficient, taking the uniformity of the large-scale fading coefficients within the cluster as an optimization constraint to ensure the uniform signal quality of the access points within the cluster, thereby improving the network stability and the accuracy of cooperative detection. This innovation makes the cluster number selection not only meet the mathematical optimality but also satisfy the reliability requirements of the physical layer, especially having significant advantages in ultra-dense deployed networks.

[0056] (4) By adding a computational complexity term to the optimization objective function and flexibly adjusting the complexity weight according to the actual requirements, the present invention can achieve a good balance between high-precision detection and low computational complexity. This means that in low-power or high-precision scenarios, the system can intelligently select an appropriate cluster number while reducing the computational complexity, thereby effectively coping with the high computational requirements in a large-scale network environment.

[0057] In summary, the present invention provides a clustering active user detection system and method based on a particle swarm optimization algorithm, which is applicable to the active user detection in a cell-free massive multiple-input multiple-output network and can effectively improve the detection accuracy and calculation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings involved in the embodiments are briefly introduced below. Obviously, the accompanying drawings in the following description are only schematic illustrations of some embodiments of the present invention, and those skilled in the art can also construct other forms of drawings based on these drawings without creative efforts.

[0059] Figure 1 It is a flowchart of the clustering active user detection method based on the particle swarm optimization algorithm proposed in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments.

[0061] The present invention proposes a clustering active user detection system and method based on a particle swarm optimization algorithm to solve the following problems:

[0062] (1) Limitations in static cluster number selection: The existing K-means algorithm relies on a fixed cluster number preset manually, which may lead to the situation that the access points within a cluster are overloaded while the resources of other clusters are idle when the network dynamic load changes. The static cluster number cannot adapt to the changes of the network. Clustering only based on geographical location during communication is difficult to ensure the communication quality in terms of communication performance, reducing the flexibility and efficiency of the system.

[0063] (2) Local optimum problem: The traditional K-means algorithm lacks a systematic consideration of communication performance in cluster number selection and is prone to falling into local optimum solutions, unable to fully optimize the actual communication conditions of the network and affecting the overall performance of the system.

[0064] (3) Problems brought by ultra-dense deployment: In the ultra-dense deployment scenario, the spatial correlation between access points and users is highly complex. Static clustering may cause the accumulation of cross-cluster interference and the decline of energy efficiency, thus affecting the scalability and intelligence level of the cell-free network in the large-scale machine-type communication connection scenario.

[0065] (4) Problem of unbalanced large-scale fading coefficients: In the existing methods, the differences in large-scale fading coefficients between access points and users are not fully considered during the cluster partitioning process, which may lead to uneven signals in some clusters, thus affecting the accuracy of cooperative detection and system performance. The following will explain the cluster active user detection system and method based on the particle swarm optimization algorithm proposed by the present invention in combination with relevant drawings and specific examples, and the specific content is as follows.

[0066] Embodiment 1:

[0067] Please refer to Figure 1 , the present invention proposes a method for detecting cluster active users based on the particle swarm optimization algorithm, which specifically includes:

[0068] First, an objective function that comprehensively considers the missed detection probability, false alarm probability, computational complexity, and variance of large-scale fading coefficients is designed. This objective function is used to evaluate the quality of cluster partitioning. The missed detection probability and false alarm probability are key indicators affecting the performance of system active user detection. The computational complexity affects the resource consumption and latency of the system. The variance of large-scale fading coefficients measures the volatility of large-scale fading coefficients within the cluster. By introducing different weight coefficients, the objective function can be flexibly adjusted to adapt to different scenario requirements, such as high-precision detection or low-power consumption scenarios. This objective function balances the detection performance and computational complexity through a multi-objective optimization framework.

[0069] Second, in order to ensure the uniform channel quality between access points and users within the cluster, the present invention introduces a constraint on the variance of large-scale fading coefficients during the optimization process. The variance of large-scale fading coefficients measures the balance of channel quality between access points and users within the cluster. A low variance indicates that the signal quality of access points within the cluster is relatively uniform, which helps to improve the accuracy of cooperative detection. If the variance of large-scale fading coefficients is large, it may lead to too large a difference in signal quality, affecting the reliability of the system. By introducing the constraint on the variance of large-scale fading coefficients, it is ensured that the large-scale fading coefficients between access points within the cluster are relatively uniform, thereby improving the cooperative detection ability of the network.

[0070] Next, the present invention uses the particle swarm optimization algorithm to dynamically optimize the selection of the number of clusters. The particle swarm optimization algorithm simulates the process of birds foraging. Each particle represents a possible number of clusters, and by continuously iterating and adjusting the position and velocity of the particles, the optimal number of clusters is found. In each iteration, the particle updates its position and velocity based on the current fitness and the global optimal position, so as to approach the global optimal solution. The position update of the particle not only takes into account the constraints of the objective function and the variance of the large-scale fading coefficient, but also ensures that the particle always searches within the physically feasible range by introducing a boundary processing mechanism. In the particle swarm optimization algorithm, the fitness function is used to evaluate the quality of the selection of the number of clusters, and the position of each particle is optimized according to the objective function and the constraints. By calculating the variance of the fitness of the particle swarm, it can be judged whether the particle swarm has converged. When the fitness variance approaches zero, it indicates that the particle swarm has found the global optimal solution and the optimization process ends.

[0071] Finally, the particle swarm optimization algorithm will output the optimal number of clusters and the corresponding fitness value. By adaptively selecting the number of clusters, the cluster division in the active user detection algorithm is optimized, which not only meets the requirements of efficient calculation, but also ensures the reliability and performance of the system. The constraint of the variance of the large-scale fading coefficient ensures the balance of the signal quality within the cluster, thereby reducing the false detection caused by excessive signal differences and improving the accuracy of active user detection.

[0072] Embodiment 2:

[0073] Based on Embodiment 1 but with differences, the present invention proposes a clustering active user detection method based on the particle swarm optimization algorithm, including the following contents:

[0074] Step 1: Initialize the particle swarm. In this stage, the position of each particle in the particle swarm represents a candidate number of clusters C, and the velocity of the particle represents the change amount of the number of clusters, which are respectively:

[0075]

[0076]

[0077] Among them, is the local optimal solution of the number of clusters, C best is the global optimal solution of the number of clusters, i represents the current index of the number of clusters, and t represents the current iteration number. When the particle swarm algorithm gives the current optimal solution of the number of clusters, it is necessary to use the fitness function to judge and decide whether to update the local optimal solution or the global optimal solution of the number of clusters according to the current fitness function value. The particle adjusts its search direction in this way and tends to find the solution that makes the fitness function value the smallest.

[0078] The initial positions of the particles are randomly generated within a preset range of the number of clusters, ensuring that the positions of all particles cover the entire possible cluster number space. For each particle, its velocity is also initialized to ensure that the particle can adjust the number of clusters according to the requirements of the objective function in subsequent iterations. The fitness value of each particle is calculated based on a multi-objective optimization function that comprehensively considers factors such as the probability of missed detection, the probability of false alarm, the computational complexity, and the variance of the large-scale fading coefficient:

[0079]

[0080] where μ1, μ2, μ3, μ4 are weight coefficients used to balance the priorities of detection performance, computational complexity, and communication quality uniformity. P md (C), P fa (C) is the probability of missed detection and the probability of false alarm, Comp(C) is the computational complexity term, is the variance of the large-scale fading coefficient.

[0081] Step 2: After the initialization of the particle swarm, the fitness value of each particle is then calculated. The fitness value is obtained through the objective function, which combines factors such as the probability of missed detection, the probability of false alarm, the computational complexity, and the variance of the large-scale fading coefficient. Each particle calculates its fitness value based on the current cluster number position, and the expression is as follows:

[0082]

[0083] where Δ max is the maximum allowable variance of the large-scale fading coefficient uniformity. η is a penalty factor used to impose an additional penalty on particles that do not satisfy the large-scale fading coefficient uniformity constraint, explicitly guiding the particles to be more inclined to satisfy the constraint. Introducing the penalty term can also explicitly penalize the cluster partitioning scheme that exceeds the uniformity threshold by the fitness function, suppressing the generation of unreasonable solutions.

[0084] Step 3: In each iteration, each particle compares its current fitness value with its historical best value. If the current fitness value is better, the personal best position of the particle is updated. The personal best position represents the best number of clusters obtained by the particle in the historical iteration process.

[0085] Step 4: The global best position is the number of clusters represented by the particle with the optimal fitness in the particle swarm. After each iteration, all particles update the global best position. If the fitness value of a certain particle is better than the current global optimal value, the global best position is updated.

[0086] Step 5: In each iteration, update the velocity and position of the particle.

[0087] Step 6: At the end of each iteration, it is necessary to check the fitness variance of the particle swarm. The fitness variance measures the fitness difference within the particle swarm. If the fitness variance is small enough, it indicates that the particle swarm has converged, and the algorithm can stop and output the global optimal number of clusters. If the fitness variance is large, it means that the particle swarm has not converged, and the particle swarm continues to iterate for optimization until the convergence condition is met, as follows:

[0088]

[0089] where I PSO represents the total number of particles in the particle swarm, which represents the total number of cluster number solutions in the problem of this article. Fitness is the fitness value of the I-th particle, and Mean(Fitness) is the average value of the particle fitness. When the variance approaches zero, it indicates that all particles in the particle swarm have approached the global optimal solution or a better solution, and the search process is completed; on the contrary, if the variance is large, it means that there is still a certain fitness difference in the particle swarm, and the search has not yet reached the final convergence.

[0090] Step 7: When the particle swarm converges, the algorithm will output the number of clusters corresponding to the global best particle, which is the optimal number of clusters C * .

[0091] In practical applications, assume that a certain cellular network operates under different loads, such as a high-density urban environment or a low-density rural environment. In the traditional method, it may be necessary to manually set different numbers of clusters for different network environments. However, through the particle swarm optimization algorithm of the present invention, the system can adaptively select the number of clusters in real time. For example, in the case of a high network load, the system will select more clusters to share the load, while in the case of a low network load, the number of clusters will be reduced, thereby reducing the computational complexity.

[0092] It should be noted that in this invention patent, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0093] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A cluster active user detection system based on the particle swarm optimization algorithm, characterized in that, It includes the following modules: K-means network function module: Using the K-means algorithm to cluster the geographical locations of access points and user devices (only clustering for access points), calculate the distance between the user and the access point, and allocate the access points to different clusters according to the distance; Covariance matrix optimization and active mode decoding function module: Used to decode the active user mode and calculate the active parameters of the user according to the received signal; By minimizing the negative log-likelihood function and optimizing the covariance matrix, gradually adjust the active state parameters of each user to accurately determine whether the user is active; Large-scale fading coefficient calculation module: Calculate the large-scale fading coefficient within each cluster according to the distance between the user and the access point, and obtain the overall large-scale fading coefficient of each cluster through the weighted average method; Signature sequence generation module: Generate a unique signature sequence for identifying each user. By generating a signature sequence with a complex Gaussian distribution, provide a unique signal identifier for the user device to ensure the accuracy of signal decoding and active state detection; Complex Gaussian noise generation module: Used to simulate the noise in the wireless channel, generate complex Gaussian noise and add it to the received signal; Miss detection and false alarm probability calculation module: Used to calculate two key performance indicators of the active user detection system, namely the miss detection probability and the false alarm probability, and adjust the detection results according to different thresholds; Constant parameter calculation module: Responsible for calculating polynomial coefficients for decoding the active state of the user; Particle swarm optimization function module: Realize dynamic optimization of cluster number selection based on the particle swarm optimization algorithm; Fitness function calculation module: Used to calculate the fitness value under a given number of clusters. The fitness value comprehensively considers multiple factors such as the miss detection probability, the false alarm probability, the computational complexity, and the variance of the large-scale fading coefficient; Large-scale fading coefficient variance calculation module: Used to calculate the variance of the large-scale fading coefficient within each cluster to measure the uniformity of the signal quality within the cluster; Main program module: Responsible for initializing parameters and performing Monte Carlo simulation, setting the range of cluster number selection according to different numbers of access points, and running the particle swarm optimization algorithm for cluster number optimization.

2. The clustering active user detection system based on the particle swarm optimization algorithm according to claim 1, wherein The specific execution process of the particle swarm optimization function module is as follows: In particle swarm optimization, represent the position of the particle as the number of clusters and the speed as the change amount of the number of clusters; each particle adjusts its position and speed according to the fitness value of the current number of clusters; The particle updates its personal best and global best solutions during the search process and updates its speed according to the inertia weight; The particle swarm optimization algorithm obtains the optimal number of clusters through multiple iterations.

3. A clustering active user detection system based on the particle swarm optimization algorithm according to claim 1, characterized in that, The specific execution process of the fitness function calculation module is as follows: For a given number of clusters, calculate the large-scale fading coefficient between the access points and users within the cluster; Calculate the variance of the large-scale fading coefficient to reflect the uniformity of the signal quality within the cluster; Calculate the miss detection and false alarm probabilities; Weight the above calculated indicators to obtain the fitness value of this number of clusters.

4. A clustering active user detection system based on the particle swarm optimization algorithm according to claim 1, characterized in that The specific execution process of the large-scale fading coefficient variance calculation module is as follows: Calculate the large-scale fading coefficient between the access points and users according to the cluster assignment, and calculate the weighted average based on the large-scale fading coefficient; Calculate the variance of the large-scale fading coefficient, which reflects the signal quality difference of access points within each cluster: If the variance of the large-scale fading coefficient of a certain cluster is small, it indicates that the signal quality of access points within the cluster is uniform, which is beneficial to cooperative detection; if the variance is large, it means that the signal quality difference of access points within the cluster is large and the detection effect is poor.

5. The cluster active user detection system based on the particle swarm optimization algorithm according to claim 1, characterized in that The specific execution process of the main program operation module is as follows: For each number of access points, the main program optimizes the number of clusters by calling the particle swarm optimization function and calculates the large-scale fading coefficient and complexity performance metrics; Through multiple experiments, accumulate the calculation results to obtain the average optimal number of clusters, computational complexity, and variance of the large-scale fading coefficient under different APs and numbers of access points; The main program generates charts to show the relationship between the number of clusters, complexity, and variance of the large-scale fading coefficient with the change in the number of access points, so as to analyze and optimize the network design.

6. A method for detecting clustered active users based on the particle swarm optimization algorithm implemented by using the system according to claim 1, characterized in that It includes the following steps: S1. Design an objective function that comprehensively considers the missed detection probability, false alarm probability, computational complexity, and variance of the large-scale fading coefficient, and use it to evaluate the quality of cluster partitioning; different weight coefficients are introduced during the evaluation process to adjust the objective function to meet the requirements of different scenarios; S2. Introduce the variance constraint of the large-scale fading coefficient into the system optimization process to ensure the uniform channel quality between access points and users within the cluster and improve the cooperative detection ability of the detection system; S3. Based on the particle swarm optimization function module, dynamically optimize the selection of the number of clusters through the particle swarm optimization algorithm, and further realize the optimization of the number of clusters selection through the fitness function calculation module and the large-scale fading coefficient variance calculation module; S4. After the optimization is completed, output the optimal number of clusters and the corresponding fitness value through the particle swarm optimization algorithm, optimize the cluster partitioning in the active user detection work, and realize the adaptive selection of the number of clusters.

7. A method for detecting clustered active users based on the particle swarm optimization algorithm according to claim 6, characterized in that The missed detection probability and false alarm probability affect the active user detection performance of the system; the computational complexity affects the resource consumption and delay of the system.

8. The method for detecting clustered active users based on the particle swarm optimization algorithm according to claim 7, characterized in that, The variance of the large-scale fading coefficient is used to measure the balance of the channel quality between access points and users within the cluster: a low variance indicates that the signal quality of access points within the cluster is uniform, which helps to improve the accuracy of cooperative detection; a high variance means a large signal quality, which affects the reliability of the system.

9. A method for detecting active users in clusters based on the particle swarm optimization algorithm according to claim 6, characterized in that, The specific content of S3 is as follows: Use the particle swarm optimization algorithm to simulate the process of birds foraging. Each particle represents a number of clusters, and continuously adjust the position and velocity of the particles through iteration to find the optimal number of clusters; During each iteration, the particle updates its position and velocity according to the current fitness and the global optimal position to approach the global optimal solution; during the position update process of the particle, consider the constraint conditions of the objective function and the variance of the large-scale fading coefficient, and introduce a boundary processing mechanism to ensure that the particle always searches within the physically feasible range; Use the fitness function to evaluate the quality of the number of clusters selection and optimize the position of each particle according to the objective function and constraint conditions; Judge whether the particle swarm has converged by calculating the variance of the particle swarm fitness; when the fitness variance is close to zero, it indicates that the particle swarm has found the global optimal solution and the optimization process ends.

Images