Network deployment method for cell-free massive MIMO system based on service coverage

Abstract

The present application is about a network deployment method for a non-cellular massive MIMO system based on service coverage, the method comprising the following steps: constructing a system model of a non-cellular massive MIMO network, the system model comprising a service coverage area, and a plurality of access points and a plurality of communication users randomly distributed in the service coverage area, each access point being connected to an MEC server, and each access point being connected to a central processing unit via a backhaul link, the service coverage area comprising a plurality of regional coverage units; for the system model, with the goal of obtaining maximum service coverage for each regional coverage unit, constructing a joint deployment problem of all access points and all MEC servers; solving the joint deployment problem using a target cascade analysis method, and completing the network deployment of all access points and all MEC servers. The present application can achieve accurate and on-demand allocation of the transmission capacity, computing capacity, and storage capacity of the non-cellular massive MIMO network.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a network deployment method for a non-cellular large-scale MIMO system based on service coverage. Background Art

[0002] The Cell-Free Massive Multiple Input Multiple Output (CFM-MIMO) system was developed to address the severe inter-cell interference and frequent handoffs inherent in traditional cellular communication networks. This system deploys a large number of access points (APs) distributed throughout its coverage area. These APs are connected to the central processing unit (CPU) via backhaul links, enabling collaboration between the APs to provide services to users.

[0003] Cell-free massive MIMO networks can combine the performance advantages of mobile edge computing (MEC) servers and cell-free massive MIMO systems, integrating communication, computing, storage, and other functions. They offer the advantages of high efficiency, reliability, low latency, and low cost, making them a key research direction in 6G networks. However, traditional cell-free massive MIMO systems assume that access points (APs) are randomly and evenly distributed across the coverage area. This assumption is idealistic and fails to fully exploit the performance advantages of flexible cell-free network deployment. Furthermore, current research on cell-free massive MIMO technology typically optimizes overall system throughput and energy efficiency, failing to account for the uneven distribution of users and mobile services across different regions. This results in wasted service capacity and energy in some areas, leading to high network construction costs and energy consumption. Research on how to achieve the precise and personalized service requirements of 6G networks through the deployment of APs and MEC in cell-free networks is still insufficient.

[0004] Therefore, it is necessary to propose a solution to improve one or more problems existing in the above-mentioned related technical solutions.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to ordinary technicians in this field. Summary of the Invention

[0006] The embodiment of the present application provides a method for deploying a non-cellular massive MIMO system network based on service coverage, the method comprising the following steps:

[0007] Constructing a system model of a non-cellular massive MIMO network, the system model including a service coverage area, and a plurality of access points and a plurality of communication users randomly distributed in the service coverage area, each of the access points being connected to an MEC server, each of the access points being connected to a central processing unit via a backhaul link, and the service coverage area including a plurality of regional coverage units;

[0008] For the system model, with the goal of obtaining maximum service coverage for each area coverage unit, construct a joint deployment problem for all the access points and all the MEC servers;

[0009] Solve the joint deployment problem using a target cascade analysis method to complete the network deployment of all the access points and all the MEC servers;

[0010] Wherein, using the target cascade analysis method to solve the joint deployment problem includes a first solution method and a second solution method.

[0011] In an exemplary embodiment of the present application, the step of constructing a system model of a non-cellular massive MIMO network includes:

[0012] Dividing a service coverage area into a plurality of area coverage units, and randomly arranging a plurality of access points and a plurality of communication users in the plurality of area coverage units;

[0013] Using meta-learning to adjust the service coverage of each of the regional coverage units, and using the adjustment results to construct the system model;

[0014] The service coverage of each of the area coverage units is the ratio of the key transmission capability of each of the area coverage units to the average service demand; the key transmission capability includes at least the total computing capability of each of the area coverage units, the total delay of each of the area coverage units, and the total transmission capacity of each of the area coverage units; the average service demand includes at least the computing power requirement of each of the area coverage units, the minimum delay of each of the area coverage units, and the average transmission capacity of all the access points in each of the area coverage units.

[0015] In an exemplary embodiment of the present application, the expression of the service coverage of the area coverage unit is:

[0016]

[0017] Among them, θ n Indicates the service coverage of the regional coverage unit, a n1 Indicates the total computing power weighting coefficient of the nth area coverage unit, a n1 ∈[0,1],Tn (z) represents the total computing power of the nth area coverage unit, Indicates the computing power requirement of the nth area coverage unit, a n2 represents the total delay weighting coefficient of the nth area coverage unit, a n2 ∈[0,1],t n (z) represents the total delay of the nth area coverage unit, Indicates the minimum delay of the nth area coverage unit, a n3 represents the total transmission capacity weighting coefficient of the nth area coverage unit, a n3 ∈[0,1],C n (z) represents the total transmission capacity of the nth area coverage unit, represents the average transmission capacity of all access points in the nth area coverage unit, a n1 +a n2 +a n3 =1, Z represents the deployment location where the access point and MEC server are jointly deployed.

[0018] In an exemplary embodiment of the present application, the total transmission capacity of the nth area coverage unit is expressed as follows:

[0019]

[0020] in, represents z l belongs to the set of L×1 real vectors, z l is a decision vector consisting of 0 and 1, z l ={z1,z2,...,z L} T , z l =1 means that the access point and MEC server are jointly deployed at the lth deployment location, z l = 0 means that no access point and MEC server are deployed at the lth deployment location, L represents the number of deployment locations, l∈L, γ k represents the signal-to-interference-and-noise ratio of the kth communication user, W represents the system bandwidth, and U n represents the nth area coverage unit, k represents the kth communication user, k∈U n Indicates that the kth communication user is located in the nth area coverage unit, k' indicates the k'th communication user, represents the conjugate transpose of the estimated channel, g lk represents the channel between the lth access point and the kth communication user, β lk represents the large-scale fading coefficient, h lk represents the small-scale fading coefficient, represents the estimated channel, glk' represents the channel between the lth access point and the k'th communication user, p k represents the transmission power of the kth communication user, η k represents the power control coefficient of the kth communication user, represents the maximum uplink power of the kth communication user, p k' represents the transmission power of the k'th communication user.

[0021] In an exemplary embodiment of the present application, the total delay of the nth area coverage unit is expressed as:

[0022]

[0023] Among them, t n (z) represents the total delay of the nth area coverage unit, Indicates the computational delay of the MEC server in the nth area coverage unit in processing the communication user's business, represents the transmission delay of the communication user unloading data to the MEC server in the nth area coverage unit, m represents the mth MEC server, m∈U n represents the mth MEC server in the nth area coverage unit, f mk represents the processing speed of the mth MEC server executing the kth communication user, in cycles / s, and f mk ≤f max , f max Indicates the upper limit of the processing capacity of each MEC server, N cpb Indicates the number of cycles required to process 1 bit of information, in cycles / bit, T k represents the task size of the kth communication user.

[0024] In an exemplary embodiment of the present application, the step of adjusting the service coverage of each of the area coverage units using meta-learning and constructing the system model using the adjustment results includes:

[0025] Obtaining service coverage of the service coverage area by using meta-learning, and dividing the obtained service coverage of the service coverage area into a training sample set and a test sample set;

[0026] Using the training sample set to train each of the area coverage units respectively, to obtain the key transmission capability and the average service demand of each of the area coverage units;

[0027] According to the obtained key transmission capability and the average service demand of each of the area coverage units, each of the area coverage units is verified and adjusted using the test sample set to construct the system model.

[0028] In an exemplary embodiment of the present application, the initial form of the joint deployment problem is:

[0029] P1:

[0030] stC1:z l =0 or 1, l=1,...,L

[0031] C2:

[0032] Where P1 represents the initial form of the joint deployment problem, represents the initial objective function of the joint deployment problem, n represents the nth regional coverage unit, n∈N, st represents the constraint, σ represents the first redundancy factor, σ is a constant greater than 1, C1 represents constraint condition 1, C2 represents constraint condition 2, t n Indicates the minimum delay of the nth area coverage unit.

[0033] In an exemplary embodiment of the present application, the first solution includes:

[0034] The initial form P1 of the joint deployment problem is a non-convex problem. The communication capacity and computing capacity within the service coverage area are transformed into the constraints of the initial form P1 of the joint deployment problem to form problem P2. The expression of problem P2 is:

[0035] P2:

[0036] stC1a:

[0037] C1b:

[0038] C2:

[0039] C3:

[0040] C4:

[0041] C5:

[0042] in, represents the lower bound of the computational delay, represents the lower bound of the transmission delay, Indicates that n can take any value in the range [1, N], C1a indicates constraint 1a, C1b indicates constraint 1b, C3 indicates constraint 3, C4 indicates constraint 4, and C5 indicates constraint 5. Indicates that the function of the three variables in the brackets is equivalent to the expression on the right side of the equal sign. represents the objective function of problem P2;

[0043] Using Lagrangian relaxation method and penalty function method, the problem P2 is transformed into a continuous non-convex problem P3;

[0044] The expression of the continuous non-convex problem P3 is:

[0045] P3:

[0046] stC1b:

[0047] C2:

[0048] C3:

[0049] C4:

[0050] C5:

[0051] in, represents the objective function of the continuous non-convex problem P3, μ represents the penalty factor;

[0052] An auxiliary variable r is introduced, and the continuous non-convex problem P3 is transformed into a continuous convex problem P4 using a continuous convex approximation algorithm. The expression of the continuous convex problem P4 is:

[0053]

[0054] stC1b:

[0055] C5:

[0056] C6:

[0057] C7:

[0058] C8:

[0059] C9:

[0060] C10:

[0061] Among them, the feasible region of min r is represents the objective function of the continuous convex problem P4, C6 represents constraint 6, C7 represents constraint 7, C8 represents constraint 8, C9 represents constraint 9, C10 represents constraint 10, f n Indicates the computing power of the nth area coverage unit, represents f n The approximate function at the feasible solution, g n represents the throughput of the nth area coverage unit, Indicates g n The approximate function at the feasible solution, Indicates g l,1 Approximate function at a feasible solution;

[0062] The continuous convex problem P4 is iteratively solved using an interior point method or a CVX optimization tool until the solution converges to obtain an optimal solution or an infeasible solution for each of the area coverage units.

[0063] In an exemplary embodiment of the present application, the second solution includes:

[0064] After obtaining the initial form of the joint deployment problem, constructing an access point optimization deployment problem;

[0065] Solving the access point optimization deployment problem using a mean square error criterion to obtain an optimization result of the access point optimization deployment problem;

[0066] Build MEC server optimization deployment issues;

[0067] Solve the MEC server optimization deployment problem by using the optimization result of the access point optimization deployment problem to obtain the optimization result of the MEC server optimization deployment problem;

[0068] All steps of the second solution method are iterated until the objective function of the joint deployment problem converges, and the network deployment of all the access points and all the MEC servers is completed.

[0069] In an exemplary embodiment of the present application, the expression of the access point optimization deployment problem is:

[0070]

[0071] stC1':x l =0or 1,l=1,...,L

[0072] C2': C n (x)≥C min,n=1,...,N

[0073] in, represents the objective function of the access point optimization deployment problem, C1' represents constraint 1', C2' represents constraint 2', and J n (x) represents the service coverage of the access point of each area coverage unit, C n (x) represents the total transmission capacity of the nth area coverage unit in the second solution, represents a decision vector consisting of 0 and 1, x l ={x1,x2,...,x L} T , x l =1 means the access point is deployed at the lth deployment location, x l =0 means no access point is deployed at the lth deployment location, T means transposition, β lk represents the large-scale fading coefficient between the kth communication user and the lth access point;

[0074] The expression of the MEC server optimization deployment problem is:

[0075]

[0076] stC1:z l =0 or 1, l=1,...,L

[0077] C11:T n (z)≤C n (x)

[0078] C12:t comm +t comp ≤t max

[0079] C13:

[0080] Where σ' represents the second redundancy factor, t comm represents the computation delay, t comp represents the communication delay, t max Indicates the minimum delay allowed by business requirements, f n AP represents the computing resources provided by all access points within the nth area coverage unit, f CPU Indicates the computing power resources provided by the central processing unit, It represents the computing power resources allocated by the central processing unit to the kth communication user, in cycles / s. It represents the computing power resources allocated by the lth access point to the kth communication user, in cycles / s, k∈A n It indicates that the kth communication user is located in the nth area coverage unit.

[0081] This application proposes a network deployment method for a cellular-free massive MIMO system based on service coverage. The method first constructs a system model of a cellular-free massive MIMO network, and then constructs a joint deployment problem of access points and MEC servers with the goal of obtaining maximum service coverage for each regional coverage unit. Finally, the target cascade analysis method is used to solve the joint deployment problem, and the network deployment of all access points and all MEC servers in the system model is completed. This application can realize the precise and on-demand allocation of the transmission capacity, computing capacity and storage capacity of the cellular-free massive MIMO network. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0083] Figure 1 A schematic diagram illustrating the steps of a method for deploying a non-cellular massive MIMO system network based on service coverage in an exemplary embodiment of the present application;

[0084] Figure 2 A schematic diagram illustrating a system model of a non-cellular massive MIMO network in an exemplary embodiment of the present application is shown;

[0085] Figure 3 A flowchart illustrating a system model for constructing a non-cellular massive MIMO network in an exemplary embodiment of the present application;

[0086] Figure 4 A flowchart showing a method for solving the joint deployment problem using the target cascade analysis method in an exemplary embodiment of the present application is shown;

[0087] Figure 5 A comparative graph showing the minimum delay distribution of two methods, one based on random deployment and the other based on service coverage, in a simulation experiment of an exemplary embodiment of the present application is shown;

[0088] Figure 6 A schematic diagram showing a comparison of the total transmission capacity using a method based on non-uniform business demand, a method based on random deployment, and a joint deployment method based on service coverage in a simulation experiment of an exemplary embodiment of the present application;

[0089] Figure 7 A comparative curve chart showing the communication success rates of two methods, namely, random deployment and joint deployment based on service coverage, in a simulation experiment of an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0090] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0091] In addition, the accompanying drawings are merely schematic illustrations of the present application and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the blocks shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0092] This exemplary embodiment provides a method for deploying a non-cellular massive MIMO system network based on service coverage, such as Figure 1 As shown, the following steps may be included:

[0093] Step S101: Construct a system model of a non-cellular massive MIMO network, which includes a service coverage area, and multiple access points and multiple communication users randomly distributed in the service coverage area. Each access point is connected to an MEC server, and each access point is connected to a central processing unit through a backhaul link. The service coverage area includes multiple area coverage units.

[0094] Step S102: For the system model, with the goal of obtaining maximum service coverage for each area coverage unit, a joint deployment problem of all access points and all MEC servers is constructed.

[0095] Step S103: Use the target cascade analysis method to solve the joint deployment problem and complete the network deployment of all access points and all MEC servers.

[0096] Here, solving the joint deployment problem using the target cascade analysis method includes a first solving method and a second solving method.

[0097] A method for network deployment of a non-cellular massive MIMO system based on service coverage is proposed in an embodiment of the present application. The method first constructs a system model of a non-cellular massive MIMO network, and then constructs a joint deployment problem of access points and MEC servers with the goal of obtaining maximum service coverage for each regional coverage unit. Finally, the target cascade analysis method is used to solve the joint deployment problem, and the network deployment of all access points and all MEC servers in the system model is completed. The present application can realize the precise and on-demand allocation of the transmission capacity, computing capacity and storage capacity of the non-cellular massive MIMO network.

[0098] Below, each step of the above method in this exemplary embodiment will be described in more detail.

[0099] In step S101, Figure 2 As shown in Figure 1, first, a system model of a non-cellular massive MIMO network is constructed. The system model includes a defined service coverage area, which is composed of multiple area coverage units. Multiple communication users and multiple randomly arranged access points are randomly included in the service coverage area. Each access point is connected to an independent Mobile Edge Computing server (MEC), i.e., the MEC server, and is connected to the central processing unit through a backhaul link, i.e., Figure 2 The CPU in. Figure 2 It can be seen that according to the service demands of multiple communication users in the service coverage area, the service coverage area includes a high service demand area High, a medium service demand area Medium, and a low service demand area Low.

[0100] It should be noted that in this embodiment, service coverage is defined as the ratio of the critical transmission capability of each coverage area to the average service demand. The critical transmission capability includes at least the total computing power, total latency, and total transmission capacity of each coverage area; the average service demand includes at least the computing power requirement, minimum latency, and average transmission capacity of all access points within each coverage area.

[0101] Furthermore, in this embodiment, step S101 includes the following sub-steps:

[0102] Sub-step S1011: Divide the service coverage area into a plurality of area coverage units, and randomly arrange a plurality of access points and a plurality of communication users in the plurality of area coverage units.

[0103] In the sub-step S1011, the expression of the service coverage of the area coverage unit is:

[0104]

[0105] Among them, θ n Indicates the service coverage of the regional coverage unit, a n1 Indicates the total computing power weighting coefficient of the nth area coverage unit, a n1 ∈[0,1],T n (z) represents the total computing power of the nth area coverage unit, Indicates the computing power requirement of the nth area coverage unit, a n2 represents the total delay weighting coefficient of the nth area coverage unit, a n2 ∈[0,1],t n (z) represents the total delay of the nth area coverage unit, Indicates the minimum delay of the nth area coverage unit, a n3 represents the total transmission capacity weighting coefficient of the nth area coverage unit, a n3 ∈[0,1],C n (z) represents the total transmission capacity of the nth area coverage unit, represents the average transmission capacity of all access points in the nth area coverage unit, a n1 +a n2 +a n3 =1, Z represents the deployment location where the access point and MEC server are jointly deployed.

[0106] Furthermore, in the system model of this embodiment, each access point is independently equipped with an MEC server and connected to the central processing unit via a backhaul link. Here, it can be assumed that the channel between the lth access point and the kth communication user is β lk represents the large-scale fading coefficient, h lk represents the small-scale fading coefficient, h lk ~CN(0,1). Coherence period τ c is divided into τ p pilot transmission interval and τ c -τ p Uplink data transmission interval. The kth communication user sends the pilot sequence at the same time during the coherence time. ||ψ k || 2 =1,τ p = K. Therefore, the signal received at the mth access point for: in, represents the transmit power, η k represents the power control coefficient of the kth communication user, represents the maximum uplink power of the kth communication user, xk represents the uplink data of the kth communication user, Represents a τ p dimensional additive noise, and satisfying

[0107] The kth communication user's access point is combined into the following decoded form by the central processing unit: in, represents the least squares channel estimation, represents the pilot vector received by the mth access point, assumes that the access points and MEC servers are intelligently deployed in L deployment locations, and defines the nth area coverage unit as U n , then the total transmission capacity of the nth area coverage unit is expressed as:

[0108]

[0109] in, represents z l belongs to the set of L×1 real vectors, z l is a decision vector consisting of 0 and 1, z l ={z1,z2,...,z L} T , z l =1 means that the access point and MEC server are jointly deployed at the lth deployment location, z l = 0 means that no access point and MEC server are deployed at the lth deployment location, L represents the number of deployment locations, l∈L, γ k represents the signal-to-interference-and-noise ratio of the kth communication user, W represents the system bandwidth, and U n represents the nth area coverage unit, k represents the kth communication user, k∈U n Indicates that the kth communication user is located in the nth area coverage unit, k' indicates the k'th communication user, represents the conjugate transpose of the estimated channel, g lk represents the channel between the lth access point and the kth communication user, β lk represents the large-scale fading coefficient, h lk represents the small-scale fading coefficient, represents the estimated channel, g lk' represents the channel between the lth access point and the k'th communication user, p k represents the transmission power of the kth communication user, η k represents the power control coefficient of the kth communication user, represents the maximum uplink power of the kth communication user, p k' represents the transmission power of the k'th communication user.

[0110] Furthermore, suppose the kth communication user has T k bits of computationally intensive tasks, due to the limitations of the computing power and battery capacity of communication users, cannot withstand high-intensity computing tasks. Therefore, it is assumed that the kth communication user will T k All bits tasks are offloaded to the MEC server without considering local parallel computing. When the computation is offloaded to the MEC server, the delay experienced can be decomposed into the transmission delay of the offloaded data and the computation processing delay of the MEC server. Therefore, the total delay of the nth area coverage unit is expressed as:

[0111]

[0112] Among them, t n (z) represents the total delay of the nth area coverage unit, Indicates the computational delay of the MEC server in the nth area coverage unit in processing the communication user's business, represents the transmission delay of the communication user unloading data to the MEC server in the nth area coverage unit, m represents the mth MEC server, m∈U n represents the mth MEC server in the nth area coverage unit, f mk represents the processing speed of the mth MEC server executing the kth communication user, in cycles / s, and f mk ≤f max , f max Indicates the upper limit of the processing capacity of each MEC server, N cpb Indicates the number of cycles required to process 1 bit of information, in cycles / bit, T k represents the task size of the kth communication user.

[0113] Furthermore, the total computing power of the nth area coverage unit is expressed as: in, Indicates the computing resources allocated by the CPU to the kth communication user, in cycles / s. Indicates the computing power resources allocated by the lth access point to the kth communication user, in cycles / s.

[0114] Sub-step S1012: Figure 3 As shown, meta-learning is used to adjust the service coverage of each area coverage unit, and the adjustment results are used to build a system model. Sub-step S1012 of this embodiment includes the following implementation process:

[0115] First, meta-learning is used to obtain the service coverage of the service coverage area, and the obtained service coverage of the service coverage area is divided into a training sample set and a test sample set.

[0116] Secondly, each regional coverage unit is trained separately using the training sample set to obtain the key transmission capacity and average service demand of each regional coverage unit. Figure 3 In the example, f represents a model definition, SVM stands for support vector machine, and LOSSmeta represents the loss function. By using linear classifiers to generate different service weight coefficients, we can balance the requirements of diverse service demands on the critical transmission capabilities of the system model, thereby determining the critical transmission capabilities of each regional coverage unit. Based on the service demands and critical transmission capabilities of different regional coverage units, we establish an initial system model.

[0117] Finally, based on the key transmission capabilities and average service demands of each regional coverage unit, the system model was constructed by verifying and adjusting each regional coverage unit using a test sample set. Using test samples to verify and adjust the system model can improve the accuracy, systematicity, and scientific nature of service coverage.

[0118] In step S101, in the system model of a non-cellular massive MIMO network, an MEC server is configured at each access point and CPU. This enables real-time data storage, downloading, processing, and analysis at the edge of the non-cellular massive MIMO network. Given the diverse business requirements across different application scenarios, the constructed system model must consider not only communication needs but also the degree of matching business requirements such as computing power and latency. To achieve the coordinated deployment of communication, computing, and storage in this system model, it is necessary not only to optimize the number, density, and deployment locations of access points, but also the computing power of the CPU and MEC servers, the number and capacity of MEC servers deployed, and the connection between MEC servers and access points. This ensures computing, storage, and communication capabilities that match business needs. Furthermore, the CPU dynamically allocates and schedules communication resources based on traffic load, cache, and computing requirements to match the needs of communication users, achieving coordinated optimization of communication and computing capabilities, improving resource utilization and communication computing quality.

[0119] In step S102, for the system model, with the goal of obtaining maximum service coverage for each area coverage unit, a joint deployment problem of all access points and all MEC servers is constructed.

[0120] In response to the uneven communication and computing requirements in non-cellular massive MIMO networks, this embodiment will achieve computing and communication capabilities that match the needs of each regional coverage unit through the joint optimization deployment of access points and MEC servers. The constructed system model is used to characterize the computing processing capabilities of the MEC server within the service coverage area, and is used to match the uneven computing tasks of multiple communication users in each regional coverage unit. At the same time, the deployment locations of different access points will affect the transmission delay of communication users to offload data to the MEC server. In this implementation step S102, a joint deployment problem is constructed with the goal of maximizing the service coverage of each regional coverage unit, so that each regional coverage unit obtains matching communication and computing capabilities.

[0121] Furthermore, the initial form of the joint deployment problem is:

[0122] P1:

[0123] stC1:z l =0 or 1, l=1,...,L

[0124] C2:

[0125] Where P1 represents the initial form of the joint deployment problem, represents the initial objective function of the joint deployment problem, n represents the nth regional coverage unit, n∈N, st represents the constraint, σ represents the first redundancy factor, σ is a constant greater than 1, C1 represents constraint condition 1, C2 represents constraint condition 2, t n Indicates the minimum delay of the nth area coverage unit.

[0126] In step S103, Figure 4 As shown in Figure 1, the target cascade analysis method is used to solve the joint deployment problem and complete the network deployment of all access points and all MEC servers. The following describes two solution methods (the first solution method and the second solution method) respectively.

[0127] The first solution method is used to solve the joint deployment problem. The specific process is as follows:

[0128] The joint deployment problem is a multi-objective and multi-constrained mixed integer nonlinear programming problem. The initial form P1 of the joint deployment problem is a non-convex problem. When solving it:

[0129] The first step is to transform the communication capacity and computing capacity within the service coverage area into the constraints of the initial form P1 of the joint deployment problem, forming problem P2. The expression of problem P2 is:

[0130] P2:

[0131] stC1a:

[0132] C1b:

[0133] C2:

[0134] C3:

[0135] C4:

[0136] C5:

[0137] in, represents the lower bound of the computational delay, represents the lower bound of the transmission delay, Indicates that n can take any value in the range [1, N], C1a indicates constraint 1a, C1b indicates constraint 1b, C3 indicates constraint 3, C4 indicates constraint 4, and C5 indicates constraint 5. Indicates that the function of the three variables in the brackets is equivalent to the expression on the right side of the equal sign. Denotes the objective function of problem P2.

[0138] The second step is to transform the problem P2 into a continuous non-convex problem P3 using the Lagrange relaxation method and the penalty function method. Here, the integer constraints of 0 or 1 in the constraint condition C1 are relaxed, which is equivalent to the constraint conditions C1a and constraint conditions C1b. l It is relaxed to a continuous variable with a value range of 0 to 1, and then the penalty function method is used to make the constraint C1a as part of the objective function of problem P2.

[0139] The expression of the continuous non-convex problem P3 is:

[0140] P3:

[0141] stC1b:

[0142] C2:

[0143] C3:

[0144] C4:

[0145] C5:

[0146] in, represents the objective function of the continuous non-convex problem P3, and μ represents the penalty factor.

[0147] In the third step, through relaxation, the constraints C2 to C5 and the objective function's structure of discrete variables multiplied by continuous variables are transformed into a structure of continuous variables multiplied by continuous variables, but it is still a non-convex problem. At this time, the auxiliary variable r and the new non-convex constraint C10 are introduced: The continuous non-convex problem P3 can be equivalently transformed into min r, and the feasible region of min r is For the multiplication form of non-negative convex functions, first transform the form of DC (Difference of Convex, DC) function, and then find the first-order Taylor expansion of the square terms in the function at the feasible solution, so as to obtain the approximate functions of the non-negative constraints C2 to C5, C1b and C10 at the feasible solution Ω(k). Then, according to the inequality requirements of continuous convex approximation on the approximate function, the continuous convex problem P4 is obtained.

[0148] The expression of the continuous convex problem P4 is:

[0149]

[0150] stC1b:

[0151] C5:

[0152] C6:

[0153] C7:

[0154] C8:

[0155] C9:

[0156] C10:

[0157] Among them, the feasible region of min r is represents the objective function of the continuous convex problem P4, C6 represents constraint 6, C7 represents constraint 7, C8 represents constraint 8, C9 represents constraint 9, C10 represents constraint 10, f n Indicates the computing power of the nth area coverage unit, represents f n The approximate function at the feasible solution, g n represents the throughput of the nth area coverage unit, Indicates g nThe approximate function at the feasible solution, Indicates g l,1 Approximate function at a feasible solution.

[0158] The fourth step is to solve the continuous convex problem P4 by using the interior point method or the CVX optimization tool for convex optimization modeling in MATLAB. The corresponding min r approximates the minimum value of the function under the constraint C10 and the feasible solution Ω(k). According to the continuous convex approximation, let The new feasible solution is substituted into the continuous convex problem P4 to perform the next round of convex optimization on the new approximate function, that is, the continuous convex problem P4 is iteratively solved until the solution converges, and the optimal solution or infeasible solution of each regional coverage unit is obtained.

[0159] The iterative solution algorithm based on continuous convex approximation is shown in Table 1 below:

[0160] Table 1: Iterative solution algorithm based on continuous convex approximation

[0161]

[0162]

[0163] The second solution method is used to solve the joint deployment problem. The specific process is as follows:

[0164] Step 1: After obtaining the initial form of the joint deployment problem, construct the access point optimization deployment problem; the expression of the access point optimization deployment problem is:

[0165]

[0166] stC1':x l =0or 1,l=1,...,L

[0167] C2': C n (x)≥C min ,n=1,...,N

[0168] in, represents the objective function of the access point optimization deployment problem, C1' represents constraint 1', C2' represents constraint 2', and J n (x) represents the service coverage of the access point of each area coverage unit, C n (x) represents the total transmission capacity of the nth area coverage unit in the second solution, represents a decision vector consisting of 0 and 1, x l ={x1,x2,...,xL} T , x l =1 means the access point is deployed at the lth deployment location, x l =0 means no access point is deployed at the lth deployment location, T means transposition, β lk represents the large-scale fading coefficient between the kth communication user and the lth access point.

[0169] Here, we first assume that only the demand for the transmission capacity of the regional coverage unit is considered, and define J n (x), where C n (x) = log2det(I K +ρ r G H G),I K is a k×k dimensional unit matrix, G=[g1,...,g K ] is the number of channel matrices of communication users, ρ r is the uplink transmission power of the communication user, and H represents the conjugate transpose.

[0170] The asymptotic orthogonality of the communication channels of the system model can be obtained:

[0171]

[0172] Where M represents the number of antennas, diag represents the diagonal matrix, and β Km represents the large-scale fading coefficient between the Kth communication user and the mth access point, It means that when M approaches infinity, the above formula must converge to 0. Let the deployment position l∈{1,2,...,L}, and define represents a decision vector consisting of 0 and 1, x l ={x1,x2,...,x L} T , x l =1 means the access point is deployed at the lth deployment location, x l =0 means that no access point is deployed at the lth deployment position, and T means transposition.

[0173] Since the service coverage area is divided into multiple area coverage units, the service coverage of the nth area coverage unit

[0174] Among them, β nl represents the large-scale fading coefficient between the nth area coverage unit and the lth access point, k∈A n Indicates that the kth communication user is located in the nth area coverage unit. According to J n (x) and C' n(x) It can be seen that by optimizing the deployment location and coverage of access points, the transmission capacity that matches the business needs can be achieved, and the expression (8) of the access point optimization deployment problem can be constructed.

[0175] In the second step, the mean square error criterion is used to solve the access point optimization deployment problem and obtain the optimization result of the access point optimization deployment problem.

[0176] Here, according to the Mean Squared Error (MEC) criterion, J n (x) can be expressed as: K n Represents the number of communication users within the nth unit area. Solve the problem and obtain the optimization result of the access point optimization deployment problem.

[0177] The third step is to use the optimization results of the access point optimization deployment problem to solve the MEC server optimization deployment problem and obtain the optimization results of the MEC server optimization deployment problem.

[0178] Since the computing and offloading service capabilities are limited by the quality and capacity of the communication link, the joint deployment of MEC servers and screenshot points is solved as a two-layer nested optimization problem. That is, the optimal deployment of access points is regarded as the lower-layer optimization problem, and the deployment number, deployment location and coverage of access points are optimized to obtain the transmission capacity that matches the business needs. Furthermore, the optimal deployment of MEC servers is regarded as the upper-layer optimization problem, and a two-layer optimization configuration model based on iterative optimization is established.

[0179] With the goal of achieving maximum service coverage, the expression of the MEC optimization deployment problem is constructed as follows:

[0180]

[0181] stC1:z l =0 or 1, l=1,...,L

[0182] C11:T n (z)≤C n (x)

[0183] C12:t comm +t comp ≤t max

[0184] C13:

[0185] Where σ' represents the second redundancy factor, t comm represents the computation delay, t comp represents the communication delay, tmax Indicates the minimum delay allowed by business requirements. represents the computing resources provided by all access points within the nth area coverage unit, f CPU Indicates the computing power resources provided by the central processing unit, It represents the computing power resources allocated by the central processing unit to the kth communication user, in cycles / s. It represents the computing power resources allocated by the lth access point to the kth communication user, in cycles / s, k∈A n It indicates that the kth communication user is located in the nth area coverage unit.

[0186] In the fourth step, all steps of the second solution method are iterated until the objective function of the joint deployment problem converges and the network deployment of all access points and the MEC server is completed.

[0187] The results obtained by the second solution method are consistent with those obtained by the first solution method.

[0188] Depend on Figure 4 As can be seen, the joint deployment problem is solved through a two-layer parallel optimization using the Analysis Target Cascading (ATC) method. ATC is a method for hierarchically decomposing and optimizing complex problems. Its outstanding feature is that each layer can be solved independently, meeting the optimal objectives of each layer in parallel, and then optimizing the entire system model through coupled variable iteration. The hierarchical structure of ATC matches the joint deployment problem of access points and MEC servers, making the optimization of communication and computing capabilities both independent and mutually influential. The optimal solution for each layer is obtained through two-layer parallel optimization, and then the joint optimization of communication and computing is achieved through coupled variable iteration.

[0189] In the examples of the present application, in order to further verify the effect of the method proposed in the present application, the following simulation experiments were conducted:

[0190] Simulation 1: Figure 5 As shown in Figure 2, the simulation experiments compare the average delays of random deployment and joint deployment based on service coverage.

[0191] When using a random deployment approach, the average latency achieved by each coverage unit is approximately 2.5ms. This is because the randomness of the access point deployment locations in this approach results in insufficient coverage in some coverage units, leading to traffic congestion, while other coverage units have excessive coverage. This unbalanced distribution of network resources makes it more likely that user latency will exceed the minimum latency allowed by the service.

[0192] When using a joint deployment method based on service coverage, the average achievable latency within each coverage unit is approximately 0.7ms, lower than the required average latency of 1.8ms. This is because the joint deployment method based on service coverage deploys more access points in coverage units with high service demand, shortening access and transmission latency for communication users, thereby improving system model performance and communication user experience.

[0193] Simulation 2: Figure 6 a. Figure 6 b and Figure 6 As shown in Figure c, the simulation experiment evaluates and compares the method based on non-uniform business demand, the random deployment method, and the joint deployment method based on service coverage.

[0194] Depend on Figure 6 As can be seen from a, in the case of non-uniform service area distribution, coordinates x and y represent the geographical location of the area coverage unit, and coordinate z represents the achievable throughput. The service demands of communication users are non-uniformly distributed.

[0195] Depend on Figure 6 b It can be seen that when the random deployment method is adopted, although uniform and consistent service coverage is achieved, the shortcoming is that it causes excess transmission capacity in low-traffic demand areas and obviously insufficient transmission capacity in high-traffic demand areas.

[0196] Depend on Figure 6 c It can be seen that when the joint deployment method based on service coverage is adopted, it is possible to achieve Figure 6 aThe throughput is consistent with the traffic demand, which proves that the method proposed in this application can meet the business needs required by the hotspot model area.

[0197] Simulation 3: Figure 7 As shown in Figure 2, the simulation experiment compares the communication success rate of the random deployment method and the joint deployment method based on service coverage, that is, the probability of meeting the minimum communication demand and the minimum delay within the regional coverage unit.

[0198] When the random deployment method is used, the average communication success rate is about 0.7543. Figure 7 It can be seen that the curve jitters significantly, which may cause data loss or retransmission during data transmission, thus increasing communication delay and instability.

[0199] When the joint deployment method based on service coverage is adopted, the average communication success rate is about 0.9452. Figure 7 It can be seen that the curve is smooth, and the high communication success rate area is often accompanied by a low packet loss rate, which helps to improve the quality of real-time communication applications and proves that the method proposed in this application can provide efficient and stable communication.

[0200] It should be noted that although several units of the system for action execution are mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiment of the present application, the features and functions of two or more units described above can be concretized in one unit. Conversely, the features and functions of a unit described above can be further divided into multiple units for concretization. Some or all of the units can be selected according to actual needs to achieve the purpose of the present application scheme. Those of ordinary skill in the art can understand and implement it without paying creative work.

[0201] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed herein.

Claims

1. A method for deploying a non-cellular massive MIMO system network based on service coverage, characterized in that: The following steps are involved: Constructing a system model of a non-cellular massive MIMO network, the system model including a service coverage area, and a plurality of access points and a plurality of communication users randomly distributed in the service coverage area, each of the access points being connected to an MEC server, each of the access points being connected to a central processing unit via a backhaul link, and the service coverage area including a plurality of regional coverage units; For the system model, with the goal of obtaining maximum service coverage for each area coverage unit, construct a joint deployment problem for all the access points and all the MEC servers; Solve the joint deployment problem using a target cascade analysis method to complete the network deployment of all the access points and all the MEC servers; Wherein, solving the joint deployment problem by using the target cascade analysis method includes a first solution method and a second solution method; The second solution method includes: After obtaining the initial form of the joint deployment problem, constructing an access point optimization deployment problem; Solving the access point optimization deployment problem using a mean square error criterion to obtain an optimization result of the access point optimization deployment problem; Build MEC server optimization deployment issues; Solving the MEC server optimization deployment problem by using the optimization result of the access point optimization deployment problem to obtain the optimization result of the MEC server optimization deployment problem; Iterate all steps of the second solution method until the objective function of the joint deployment problem converges and the network deployment of all the access points and all the MEC servers is completed; The expression of the access point optimization deployment problem is: s.t.C1’:x l =0or 1,l=1,...,L C2’:C n (x)≥C min ,n=1,...,N in, represents the objective function of the access point optimization deployment problem, C1' represents constraint 1', C2' represents constraint 2', θ n (x) represents the service coverage of the access point of each area coverage unit, C n (x) represents the total transmission capacity of the nth area coverage unit in the second solution, represents a decision vector consisting of 0 and 1, x l ={x1,x2,...,x L } T , x l =1 means the access point is deployed at the lth deployment location, x l =0 means no access point is deployed at the lth deployment location, T means transposition, β lk represents the large-scale fading coefficient between the kth communication user and the lth access point; The expression of the MEC server optimization deployment problem is: s.t.C1: z l = 0 or 1, l = 1, ..., L C11:T n (z)≤C n (x) C12:t comm +t comp ≤t max Where σ' represents the second redundancy factor, t comm represents the computation delay, t comp represents the communication delay, t max Indicates the minimum delay allowed by business requirements. represents the computing resources provided by all access points within the nth area coverage unit, f CPU Indicates the computing power resources provided by the central processing unit, It represents the computing power resources allocated by the central processing unit to the kth communication user, in cycles / s. It represents the computing power resources allocated by the lth access point to the kth communication user, in cycles / s, k∈A n It indicates that the kth communication user is located in the nth area coverage unit.

2. The method for network deployment of a non-cellular massive MIMO system based on service coverage according to claim 1, characterized in that: The steps of constructing a system model of a non-cellular massive MIMO network include: Dividing a service coverage area into a plurality of area coverage units, and randomly arranging a plurality of access points and a plurality of communication users in the plurality of area coverage units; Using meta-learning to adjust the service coverage of each of the regional coverage units, and using the adjustment results to construct the system model; The service coverage of each of the area coverage units is the ratio of the key transmission capability of each of the area coverage units to the average service demand; the key transmission capability includes at least the total computing capability of each of the area coverage units, the total delay of each of the area coverage units, and the total transmission capacity of each of the area coverage units; the average service demand includes at least the computing power requirement of each of the area coverage units, the minimum delay of each of the area coverage units, and the average transmission capacity of all the access points in each of the area coverage units.

3. The method for deploying a non-cellular massive MIMO system network based on service coverage according to claim 2, characterized in that: The expression of the service coverage of the area coverage unit is: Among them, θ n Indicates the service coverage of the regional coverage unit, a n1 Indicates the total computing power weighting coefficient of the nth area coverage unit, a n1 ∈[0,1],T n (z) represents the total computing power of the nth area coverage unit, Indicates the computing power requirement of the nth area coverage unit, a n2 represents the total delay weighting coefficient of the nth area coverage unit, a n2 ∈[0,1],t n (z) represents the total delay of the nth area coverage unit, Indicates the minimum delay of the nth area coverage unit, a n3 represents the total transmission capacity weighting coefficient of the nth area coverage unit, a n3 ∈[0,1],C n (z) represents the total transmission capacity of the nth area coverage unit, represents the average transmission capacity of all access points in the nth area coverage unit, a n1 +a n2 +a n3 =1, Z represents the deployment location where the access point and MEC server are jointly deployed.

4. The method for network deployment of a non-cellular massive MIMO system based on service coverage according to claim 3, characterized in that: The total transmission capacity of the nth area coverage unit is expressed as: in, represents z l belongs to the set of L×1 real vectors, z l is a decision vector consisting of 0 and 1, z l ={z1,z2,...,z L } T , z l =1 means that the access point and MEC server are jointly deployed at the deployment location, z l = 0 means that no access point and MEC server are deployed at the deployment location, L represents the number of deployment locations, l∈L, γ k represents the signal-to-interference-and-noise ratio of the kth communication user, W represents the system bandwidth, and U n represents the nth area coverage unit, k represents the kth communication user, k∈U n Indicates that the kth communication user is located in the nth area coverage unit, k' indicates the k'th communication user, represents the conjugate transpose of the estimated channel, g lk represents the channel between the lth access point and the kth communication user, β lk represents the large-scale fading coefficient, h lk represents the small-scale fading coefficient, represents the estimated channel, g lk' represents the channel between the lth access point and the k'th communication user, p k represents the transmission power of the kth communication user, η k represents the power control coefficient of the kth communication user, represents the maximum uplink power of the kth communication user, p k' represents the transmission power of the k'th communication user.

5. The method for deploying a non-cellular massive MIMO system network based on service coverage according to claim 3, characterized in that: The total delay of the nth area coverage unit is expressed as: Among them, t n (z) represents the total delay of the nth area coverage unit, Indicates the computational delay of the MEC server in the nth area coverage unit in processing the communication user's business, represents the transmission delay of the communication user unloading data to the MEC server in the nth area coverage unit, m represents the mth MEC server, m∈U n represents the mth MEC server in the nth area coverage unit, f mk represents the processing speed of the mth MEC server executing the kth communication user, in cycles / s, and f mk ≤f max , f max Indicates the upper limit of the processing capacity of each MEC server, N cpb Indicates the number of cycles required to process 1 bit of information, in cycles / bit. represents the task size of the kth communication user.

6. The method for network deployment of a non-cellular massive MIMO system based on service coverage according to claim 3, characterized in that: The step of adjusting the service coverage of each of the area coverage units by using meta-learning and constructing the system model by using the adjustment results includes: Obtaining service coverage of the service coverage area by using meta-learning, and dividing the obtained service coverage of the service coverage area into a training sample set and a test sample set; Using the training sample set to train each of the area coverage units respectively, to obtain the key transmission capability and the average service demand of each of the area coverage units; According to the obtained key transmission capability and the average service demand of each of the area coverage units, each of the area coverage units is verified and adjusted using the test sample set to construct the system model.

7. The method for network deployment of a non-cellular massive MIMO system based on service coverage according to claim 4, characterized in that: The initial form of the joint deployment problem is: s.t.C1: z l = 0 or 1, l = 1, ..., L Where P1 represents the initial form of the joint deployment problem, represents the initial objective function of the joint deployment problem, n represents the nth regional coverage unit, n∈N, st represents the constraint, σ represents the first redundancy factor, σ is a constant greater than 1, C1 represents constraint condition 1, C2 represents constraint condition 2, t n Indicates the minimum delay of the nth area coverage unit.

8. The method for network deployment of a non-cellular massive MIMO system based on service coverage according to claim 7, characterized in that: The first solution method includes: The initial form P1 of the joint deployment problem is a non-convex problem. The communication capacity and computing capacity within the service coverage area are transformed into the constraints of the initial form P1 of the joint deployment problem to form problem P2. The expression of problem P2 is: in, represents the lower bound of the computational delay, represents the lower bound of the transmission delay, Indicates that n can take any value in the range [1, N], C1a indicates constraint 1a, C1b indicates constraint 1b, C3 indicates constraint 3, C4 indicates constraint 4, and C5 indicates constraint 5. Indicates that the function of the three variables in the brackets is equivalent to the expression on the right side of the equal sign. represents the objective function of problem P2; Using Lagrangian relaxation method and penalty function method, the problem P2 is transformed into a continuous non-convex problem P3; The expression of the continuous non-convex problem P3 is: in, represents the objective function of the continuous non-convex problem P3, μ represents the penalty factor; An auxiliary variable r is introduced, and the continuous non-convex problem P3 is transformed into a continuous convex problem P4 using a continuous convex approximation algorithm. The expression of the continuous convex problem P4 is: Among them, the feasible region of min r is represents the objective function of the continuous convex problem P4, C6 represents constraint 6, C7 represents constraint 7, C8 represents constraint 8, C9 represents constraint 9, C10 represents constraint 10, f n Indicates the computing power of the nth area coverage unit, represents f n The approximate function at the feasible solution, g n represents the throughput of the nth area coverage unit, Indicates g n The approximate function at the feasible solution, Indicates g l,1 Approximate function at a feasible solution; The continuous convex problem P4 is iteratively solved using an interior point method or a CVX optimization tool until the solution converges to obtain an optimal solution or an infeasible solution for each of the area coverage units.

Images