A joint optimization method based on user association and backhaul bandwidth configuration

Abstract

The present application belongs to the field of data classification technology, and in particular, relates to a joint optimization method based on user association and backhaul bandwidth configuration. Some current models about sEMG signal recognition have low accuracy, and there are also overfitting and underfitting, gradient disappearance, poor robustness, and long training time in the training model process. The present application provides a joint optimization method based on user association and backhaul bandwidth configuration, including constructing a non-convex mixed integer fractional optimization model, de-fractionating the model and converting it into a non-convex mixed integer nonlinear optimization model, decomposing the non-convex mixed integer nonlinear optimization model into a backhaul unit bandwidth configuration optimization sub-model and a dual-connection feasible user association optimization sub-model, and alternately solving the backhaul unit bandwidth configuration optimization sub-model and the dual-connection feasible user association optimization sub-model. The optimal backhaul unit bandwidth configuration factor value can be obtained, while having the optimal system throughput and system throughput utility and performance.

Description

Technical Field

[0001] The present application belongs to the field of data classification technology, and in particular relates to a joint optimization method based on user association and backhaul bandwidth configuration. Background Art

[0002] With the rapid development of mobile communication technology, the number and variety of smart mobile terminals are constantly increasing, and people's demand for wireless network transmission speeds is increasing. To support emerging mobile communication scenarios, the fifth-generation (5G) mobile communication system features network heterogeneity, dense deployment, and diversified services. As a candidate key technology supporting 5G, dual connectivity technology has attracted widespread attention in academia.

[0003] Dual connectivity allows user devices to associate with and communicate with a macro base station and a small base station, deployed either co-frequency or inter-frequency. Fundamentally, dual connectivity is an evolved small cell enhancement technology, an implementation of carrier aggregation technology in non-ideal backhaul network scenarios. Therefore, compared to traditional single-association heterogeneous wireless networks, heterogeneous network architectures with dual connectivity effectively improve system throughput, particularly for users at the edge of the system, by increasing spectrum utilization. In a heterogeneous wireless network with dual connectivity, user devices can associate with a macro base station and a small base station, implementing dual connectivity. The small base station connects to the macro base station via a backhaul link from the macro base station to the small base station. The macro base station acts as the service network manager, handling all user data requests between the macro base station and the core network. Therefore, the performance of dual connectivity is largely limited by backhaul capacity. In heterogeneous wireless networks with non-ideal backhaul, where macro and small base stations are deployed co-frequency, the backhaul bandwidth configuration mechanism becomes a crucial model.

[0004] Existing technologies only consider heterogeneous network deployment scenarios with different frequencies, but not co-channel (co-frequency) deployment scenarios. In particular, in heterogeneous wireless networks with non-ideal backhaul where macro and small base stations are deployed on the same frequency, there is no solution to the joint optimization model for user association and backhaul bandwidth configuration that are feasible for dual connectivity. Even considering the non-ideal backhaul heterogeneous wireless network scenario where macro and small base stations are deployed on the same frequency, it only focuses on the joint optimization model for user association and fronthaul bandwidth configuration that are feasible for dual connectivity; it only considers the fixed wireless backhaul limitations from each small base station to the macro base station. Summary of the Invention

[0005] 1. Technical model to be solved

[0006] Based on some current joint optimization algorithms for sEMG signals based on user association and return bandwidth configuration, the recognition accuracy is low, and the training model process may also suffer from overfitting and underfitting, gradient vanishing, poor robustness, and long training time. This application provides a joint optimization method based on user association and return bandwidth configuration.

[0007] 2. Technical solution

[0008] To achieve the above-mentioned objectives, the present application provides a joint optimization method based on user association and backhaul bandwidth configuration. The method includes constructing a non-convex mixed integer fractional optimization model, defractionating the model into a non-convex mixed integer nonlinear optimization model, decomposing the non-convex mixed integer nonlinear optimization model into a backhaul unit bandwidth configuration optimization sub-model and a dual-connectivity feasible user association optimization sub-model, and alternately solving the backhaul unit bandwidth configuration optimization sub-model and the dual-connectivity feasible user association optimization sub-model.

[0009] Another implementation manner provided by the present application is that the non-convex mixed integer fraction optimization model includes a continuous backhaul unit bandwidth configuration factor variable and a binary base station-user association variable.

[0010] Another implementation method provided by the present application is: the alternating solution adopts an iterative update optimization algorithm based on alternating optimization.

[0011] Another implementation manner provided by the present application is that: the backhaul unit bandwidth configuration optimization sub-model is based on fixed user association, and the dual-connectivity feasible user association optimization sub-model is based on a fixed backhaul bandwidth configuration factor.

[0012] Another implementation provided by the present application is that the non-convex mixed integer fraction optimization model maximizes the sum of user throughput utilities while taking into account user rate requirements by jointly optimizing user association variables and fronthaul bandwidth configuration variables.

[0013] Another implementation provided by the present application is: the backhaul unit bandwidth configuration optimization submodel is a backhaul resource configuration optimization model with known user association, and the dual-connection feasible user association optimization submodel is a user association optimization submodel with known backhaul resource configuration factors.

[0014] Another implementation provided by the present application is that the alternating solution includes obtaining the optimal backhaul bandwidth configuration mechanism by fixing user association variables; and solving a dual-connection feasible user association sub-sub-model by fixing backhaul bandwidth configuration factor variables.

[0015] Another implementation method provided by the present application is: it also includes using a mathematical simulation method to verify the effectiveness and superiority of the joint optimization method based on user association and backhaul bandwidth configuration.

[0016] Another implementation method provided by the present application is: there are fractional terms in the optimization objectives and constraints of the non-convex mixed integer fractional optimization model, there are continuous variables and discrete variables, and there is a coupled product relationship between the continuous variables and the discrete variables in the optimization objectives and several constraints.

[0017] Another implementation method provided by the present application is: the method is based on a downlink transmission process of a two-layer heterogeneous network, and the network includes a macro base station and several small base stations with open access.

[0018] 3. Beneficial effects

[0019] Compared with the existing technology, the joint optimization method based on user association and backhaul bandwidth configuration provided by this application has the following advantages:

[0020] The joint optimization method based on user association and backhaul bandwidth configuration provided in this application is a new joint optimization method for user association and backhaul bandwidth configuration in heterogeneous wireless networks with feasible dual connections. This method takes into account flexible user association mechanisms including dual connections and backhaul capacity constraints.

[0021] This application proposes a joint optimization method for user association and backhaul bandwidth configuration. Under the constraints of small cell backhaul capacity, a joint optimization model for user association and backhaul bandwidth resource configuration is established for heterogeneous wireless networks with feasible dual connectivity, with the goal of maximizing the sum of throughput utilities. This joint optimization model is modeled as a mixed integer fractional programming model.

[0022] This application proposes a joint optimization method for user association and backhaul bandwidth configuration, based on an iterative update optimization algorithm based on alternating optimization. By performing an equivalent transformation on the optimization objective and defractionating the constraints, the formulated optimization model is transformed into a mixed-integer nonlinear optimization model without a fractional structure. This non-convex optimization model is decomposed into two optimization submodels—the backhaul bandwidth configuration submodel and the user association submodel—and solved separately.

[0023] The joint optimization method based on user association and backhaul bandwidth configuration provided in this application constructs a new framework for maximizing network throughput utility and maximization.

[0024] The joint optimization method based on user association and backhaul bandwidth configuration provided in this application can obtain the optimal backhaul unit bandwidth configuration factor value compared to the fixed backhaul bandwidth configuration mechanism, while having the optimal system throughput, system throughput utility and performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1This is a schematic diagram of a downlink two-layer heterogeneous network scenario in which dual connectivity is feasible in this application;

[0026] Figure 2 This is a schematic diagram comparing the effects of the three mechanisms under different numbers of users in this application;

[0027] Figure 3 This is a diagram of the network throughput of the three mechanisms under different numbers of users in this application;

[0028] Figure 4 This is a schematic diagram of the backhaul unit bandwidth configuration factors of the three mechanisms under different numbers of users in this application;

[0029] Figure 5 This is a diagram of the business request acceptance rate under different microservice instance specifications of this application;

[0030] Figure 6 This is a diagram of the network throughput utility of the three mechanisms under different user minimum rate requirements of this application;

[0031] Figure 7 This is a schematic diagram of the backhaul unit bandwidth configuration factors of the three mechanisms under different user minimum rate requirements of this application. DETAILED DESCRIPTION

[0032] Hereinafter, specific embodiments of the present application will be described in detail with reference to the accompanying drawings. Based on these detailed descriptions, those skilled in the art will be able to clearly understand the present application and implement the present application. Without violating the principles of the present application, the features of different embodiments may be combined to obtain new implementations, or certain features of certain embodiments may be substituted to obtain other preferred implementations.

[0033] See also Figures 1 to 7 The present application provides a joint optimization method based on user association and backhaul bandwidth configuration, the method comprising constructing a non-convex mixed integer fractional optimization model, defractionating the model into a non-convex mixed integer nonlinear optimization model, decomposing the non-convex mixed integer nonlinear optimization model into a backhaul unit bandwidth configuration optimization sub-model and a dual-connection feasible user association optimization sub-model, and alternately solving the backhaul unit bandwidth configuration optimization sub-model and the dual-connection feasible user association optimization sub-model.

[0034] This application considers the downlink transmission process of a two-layer heterogeneous network. The network consists of a macro base station and multiple openly accessible small base stations, which jointly serve randomly distributed users within the network coverage area. First, it is assumed that all user devices have two radio interfaces and are capable of communicating with a macro base station and a small base station at the same time. Next, it is assumed that the wireless backhaul link from the small base station to the macro base station and the fronthaul link from the small base station to the user device are deployed in a co-channel manner. Each user device is allocated several sub-channels, and it is assumed that all channels are smoothly fading, and each small base station allocates the same power to each sub-channel.

[0035] like Figure 1 As shown in the figure, this application considers a downlink heterogeneous wireless network scenario consisting of a macro base station, M small base stations and N randomly distributed users. The set of all users, the set of all small base stations and the set of all base stations are defined as and Therefore, the above sets can be expressed as and in Element 0 represents a macro base station, and there are

[0036] Small base stations are equipped with a single antenna and use a single-input single-output (SISO) mode for signal transmission. Macro base stations deploy millimeter-wave antenna arrays (number of antennas A t ), the beam-forming group size for small base stations and macro users is A sbs , A ue (A t >>A sbs , A t >>A ue ).

[0037] Furthermore, the non-convex mixed integer fraction optimization model includes a continuous backhaul unit bandwidth configuration factor variable and a binary base station-user association variable.

[0038] Furthermore, the alternating solution adopts an iterative update optimization algorithm based on alternating optimization.

[0039] Furthermore, the backhaul unit bandwidth configuration optimization sub-model is based on fixed user association, and the dual-connectivity feasible user association optimization sub-model is based on a fixed backhaul bandwidth configuration factor.

[0040] Furthermore, the non-convex mixed integer fraction optimization model maximizes the sum of user throughput utilities while considering user rate requirements by jointly optimizing user association variables and fronthaul bandwidth configuration variables.

[0041] Furthermore, the backhaul unit bandwidth configuration optimization sub-model is a backhaul resource configuration optimization model when user association is known, and the dual-connection feasible user association optimization sub-model is a user association optimization sub-model when backhaul resource configuration factors are known.

[0042] Furthermore, the alternating solution includes obtaining an optimal backhaul bandwidth configuration mechanism by fixing user association variables; and solving a feasible user association sub-sub-model for dual connectivity by fixing backhaul bandwidth configuration factor variables.

[0043] Assume that Orthogonal Frequency Division Multiple Access (OFDMA) is used. α is defined as the backhaul unit bandwidth configuration factor for each small base station, 0≤α≤1.

[0044] Assuming that user equipment i is associated with the jth small base station, the maximum user rate per unit bandwidth that can be obtained can be expressed as:

[0045]

[0046] Among them, p i，j is the transmission power of the signal sent by the j-th small base station to the user equipment i, g i，j is the path gain of the signal sent from the j-th small base station to the user equipment i, is the interference of the macro base station to the user equipment i associated with the j-th small base station, σ 2 is the additive white Gaussian noise power.

[0047] The maximum backhaul rate per unit bandwidth of small cell j can be expressed as:

[0048]

[0049] Assuming that user equipment i is associated with a macro base station, the maximum user rate per unit bandwidth that it can obtain can be expressed as:

[0050]

[0051]

[0052] Assume that the user association matrix is z, Then z i，j The following definitions exist:

[0053]

[0054] According to the dual connectivity mechanism considered in this application, the user equipment can be associated with only a macro base station or a small base station, or can be associated with both a macro base station and a small base station at the same time. Then, the association constraints can be obtained as follows:

[0055]

[0056]

[0057] Assume that each base station allocates uniform forward bandwidth to users associated with it. If the number of users associated with base station j is u j , then the unit bandwidth allocated by base station j to user equipment i is Then the long-term rate provided by base station j to user equipment i can actually be expressed as If the total throughput available to user equipment i is R i , which can be expressed as:

[0058]

[0059] Then the user's rate constraint can be written as follows:

[0060]

[0061] in is the minimum rate required by user i. In addition, considering that the fronthaul capacity of each small base station cannot exceed its obtained backhaul capacity, the following constraints must be met:

[0062]

[0063] The optimization goal is to maximize the sum of user throughput utilities while taking into account user rate requirements by jointly optimizing user association variables and fronthaul bandwidth configuration variables. The final optimization model can be modeled as follows:

[0064] (P1)

[0065] st(C1)

[0066] (C2)

[0067] (C3)

[0068] (C4)

[0069] (C5)

[0070] (C6)α∈[0,1]

[0071] (C7)

[0072] (C8)

[0073] in α is a continuous single variable. In the P1 model, the objective function represents maximizing the sum of user throughput utilities. The first constraint ensures that the fronthaul capacity of each small cell cannot exceed its available backhaul capacity. The second constraint ensures that the minimum required rate for each user is met. Constraints 3, 4, and 5 jointly ensure that each user terminal can be simultaneously associated with a macro cell and a small cell, implementing a dual-connection association mechanism. Constraints 5 and 6 are, respectively, binary user association variable constraints and continuous unit backhaul bandwidth resource configuration variable constraints allocated to each small cell.

[0074] Through observation, the P1 model has the following three significant characteristics: First, there are fractional terms in the optimization objectives and constraints. Second, the two optimization variables include both continuous variables α and discrete variables z. i，j ; Third, there is a coupled product relationship between continuous variables and discrete variables in the optimization objectives and several constraints, such as αz i，j Therefore, the P1 model is a non-convex mixed integer fraction optimization model. Obviously, this model is also an NP-hard model. At the same time, as the model size increases, the solution complexity increases. Usually, this type of model cannot be solved by an exact algorithm, and an effective approximate algorithm must be sought.

[0075] To simplify the P1 model, we first consider removing the fractions in the optimization objective and constraints. The simplified form is as follows:

[0076] (P2)

[0077] st(C1)

[0078] (C2)

[0079] (C3)

[0080] (C4)

[0081] (C5)

[0082] (C6)α∈[0,1]

[0083] (C7)

[0084] (C8)

[0085] (C9)

[0086] Among them, the optimization goal is to make an equivalent transformation with the help of the properties of the log function; in the first constraint, due to represents the number of users associated with base station j. Therefore, multiplying both sides of constraint 1 by this term can yield the first constraint of P2. In the second constraint, due to the difference in the sum, constraint 2 cannot be equivalently converted like constraint 1. To simplify the model, we use the constant K = N / M instead. This means that N users are evenly associated with M base stations. It should be emphasized that this simplifying assumption is reasonable based on the assumptions of random user deployment and uniform distribution of forward resources, as well as the natural fairness property of the log function.

[0087] The converted P2 model remains a mixed-integer nonlinear optimization model. Its non-convex nature remains unchanged, and it remains NP-hard. For this optimization model, we employ an alternating optimization approach to decompose the P2 model into two optimization sub-models: a backhaul resource configuration optimization model with known user associations, and a user association optimization model with known backhaul resource configuration factors.

[0088] 3) Model solution

[0089] First, given known user associations, the P2 model can be transformed into a backhaul resource configuration optimization model, as shown below:

[0090] (P2-1)

[0091] st(C1)

[0092] (C2)

[0093] (C6)α∈[0,1)

[0094] It can be observed that the P2-1 sub-model is a continuous variable convex optimization model, which is obviously easy to solve. Through transformation, the C1 constraint can be equivalent to the following form:

[0095] α≥α1

[0096]

[0097] Through transformation, the C2 constraint can be equivalent to the following form:

[0098] α≤α2

[0099]

[0100] In summary, if α1>α2, the P2-1 model has no solution, which means that the basic condition of the P2-1 model, that is, the user association solution is fixed and known, is not reasonable; if α1≤α2, the solution of the P2-1 model is α=α1.

[0101] Then, when the backhaul resource configuration factors are known, the P2 model can be transformed into a joint optimization model for user association and fronthaul resource configuration, as shown below:

[0102] (P2-2)

[0103] st(C1)

[0104] (C2)

[0105] (C3)

[0106] (C4)

[0107] (C5)

[0108] (C6)

[0109] (C7)

[0110] With the help of load variables And by introducing three Lagrange multipliers λ, v and ω, the Lagrange function corresponding to model P2-2 can be written as follows:

[0111]

[0112] After transposition and classification, the Lagrangian function can be further written as follows:

[0113]

[0114] in: and They are expressed in the following forms:

[0115]

[0116]

[0117] Through observation, we can find and It is decoupled. Therefore, according to the Langrange duality method, solving P2-2 can be decomposed into two sub-models, that is, solving and The maximum value of . Then the solution z of the first sub-model can be defined as follows:

[0118]

[0119] in

[0120]

[0121] Similarly, by setting The partial derivative of is 0. Under the condition of satisfying the C7 constraint, we can get μ j Value:

[0122]

[0123] For the update of Lagrange multipliers λ, v and ω, this application adopts the traditional sub-gradient method. The specific update process is as follows:

[0124]

[0125]

[0126]

[0127] Where [a] + =max(a,0), ε1, ε2 and ε3 are three appropriately selected step values.

[0128] With the help of (5-16) and (5-17), the corresponding variable solutions are obtained, and several Lagrange multipliers are updated at the same time. As these three multipliers are updated, iterated and finally converged, the P2-2 model can be solved.

[0129] 4) Joint optimization algorithm for dual association and backhaul bandwidth resource allocation

[0130] Based on the solution process of the above two sub-optimization models, the description of the solution algorithm of the final model P1 is summarized in Table 1 below.

[0131] Table 1 Proposed JDCBA algorithm

[0132]

[0133]

[0134] Furthermore, it also includes using a mathematical simulation method to verify the effectiveness and superiority of the joint optimization method based on user association and backhaul bandwidth configuration.

[0135] Furthermore, there are fractional terms, continuous variables and discrete variables in the optimization objectives and constraints of the non-convex mixed integer fractional optimization model, and there is a coupled product relationship between the continuous variables and the discrete variables in the optimization objectives and several constraints.

[0136] Furthermore, the method is based on a downlink transmission process of a two-layer heterogeneous network, wherein the network includes a macro base station and several small base stations with open access.

[0137] This application compares the performance of the proposed dual-connection and backhaul bandwidth resource allocation joint optimization algorithm (abbreviated as JDCBA algorithm) with two dual-connection methods based on fixed backhaul unit bandwidth configuration factors.

[0138] 1) Performance comparison under different numbers of users

[0139] exist Figure 2 In the , as the total number of users in the network gradually increases, the network throughput utility and performance of the three schemes show a rapid decline, and the difference between the network throughput utility and performance of each scheme becomes larger. Numerical calculations show that the network throughput utility and performance of the JDCBA scheme proposed in this application are 7.85% and 27.70% higher than those of the DC-α1 scheme and DC-α2 scheme on average. Figure 3 As the total number of users in the network gradually increases, the network throughput performance of the proposed JDCBA solution shows a slow growth trend, while the network throughput performance of the other two solutions remains unchanged. Numerical calculations show that in terms of network throughput performance, the proposed JDCBA solution outperforms the DC-α1 solution and the DC-α2 solution by an average of 22.03% and 144.06%, respectively.

[0140] 1) Performance comparison under different user minimum required rates

[0141] By observation Figure 5 It can be seen that the proposed JDCBA scheme has the best system throughput utility performance, followed by the DC-α1 scheme, and the DC-α2 scheme has the worst system throughput utility performance. Numerical calculations show that the proposed JDCBA scheme's system throughput utility performance is, on average, 9.51% higher than that of the DC-α1 and DC-α2 schemes, respectively.

[0142] Through observation, it was found that Figure 6 and Figure 5Similarly, the proposed JDCBA scheme has the best system throughput performance, followed by the DC-α1 scheme, and the DC-α2 scheme has the worst system throughput performance. Numerical calculations show that the proposed JDCBA scheme's system throughput performance is, on average, 20.62% higher than the DC-α1 and DC-α2 schemes, respectively, by 140.20%.

[0143] This application compares the performance of the proposed algorithm with two dual-connectivity methods based on fixed backhaul unit bandwidth configuration factors. These two comparison methods are referred to as "DC-α1" and "DC-α2," where α1 = 0.4 and α2 = 0.7. It is important to emphasize that all simulation results are obtained using over 1,000 Monte Carlo averages.

[0144] Figure 2 As the total number of users in the network gradually increases, the network throughput utility and performance of the three schemes show a rapid decline, and the difference in network throughput utility and performance between the two schemes increases. Numerical calculations show that the network throughput utility and performance of the proposed JDCBA scheme are 7.85% and 27.70% higher than those of the DC-α1 and DC-α2 schemes, respectively, on average.

[0145] Figure 3 As the total number of users in the network gradually increases, the network throughput performance of the proposed JDCBA solution shows a slow growth trend, while the network throughput performance of the other two solutions remains unchanged. Numerical calculations show that in terms of network throughput performance, the proposed JDCBA solution outperforms the DC-α1 solution and the DC-α2 solution by an average of 22.03% and 144.06%, respectively.

[0146] Figure 4 Compared with the DC-α1 and DC-α2 solutions, the JDCBA solution proposed in this application has the smallest backhaul unit bandwidth configuration factor value. Figure 4 In the figure, as the total number of users in the network gradually increases, the backhaul unit bandwidth configuration factor value of the proposed JDCBA scheme shows a slowly decreasing trend.

[0147] Figure 5 This is the business request acceptance rate under different microservice instance specifications. Since the business request acceptance rate is inversely proportional to resource fragmentation, when the microservice instance specification is larger, there is more resource fragmentation, resulting in a lower business request acceptance rate; conversely, when the microservice instance specification is smaller, there is less resource fragmentation and a higher business request acceptance rate.

[0148] Through observation, it was found that Figure 6 and Figure 5Similarly, the JDCBA scheme proposed in this application has the best system throughput performance, followed by the DC-α1 scheme, and the DC-α2 scheme has the worst system throughput performance. Numerical calculations show that the system throughput performance of the JDCBA scheme proposed in this application is 20.62% and 140.20% higher than the system throughput utility performance of the DC-α1 scheme and the DC-α2 scheme, respectively. Moreover, in Figure 6 In the figure, the three curves are horizontal and parallel. This indicates that as user rate requirements in the network gradually increase, the proposed JDCBA scheme, like the two schemes based on fixed backhaul unit bandwidth configuration factors, shows no observable change in system throughput utility and performance.

[0149] Through observation, it was found that Figure 7 and Figure 5 、 Figure 6 Similarly, compared with the DC-α1 and DC-α2 schemes, the JDCBA scheme proposed in this application has the smallest backhaul unit bandwidth configuration factor value, that is, value. Moreover, Figure 5 and Figure 6 Similarly, as the user rate requirements in the network gradually increase, the value of the proposed JDCBA scheme does not change observably.

[0150] Although the present application has been described above with reference to specific embodiments, it should be understood by those skilled in the art that many modifications may be made to the configurations and details disclosed herein within the principles and scope of the present application. The scope of protection of the present application is determined by the appended claims, and the claims are intended to cover all modifications encompassed by the literal meaning or scope of equivalents of the technical features in the claims.

Claims

1. A joint optimization method based on user association and backhaul bandwidth configuration, characterized by: The method includes constructing a non-convex mixed integer fractional optimization model, expressed as: (C6)α∈[0,1] in, α is a continuous single variable, is the set of all users, is the set of all small base stations; is the set of all base stations; R ij is the maximum user rate per unit bandwidth when user equipment i is associated with the jth small base station; is the maximum backhaul rate per unit bandwidth of small cell j; is the minimum rate required by user i; u j is the number of users associated with base station j; The model is defractionated and converted into a non-convex mixed integer nonlinear optimization model, which is expressed as: (C6)α∈[0,1] Where K is a constant, indicating that N users are evenly associated with M base stations; The non-convex mixed integer nonlinear optimization model is decomposed into a backhaul unit bandwidth configuration optimization sub-model and a dual-connectivity feasible user association optimization sub-model, and the backhaul unit bandwidth configuration optimization sub-model and the dual-connectivity feasible user association optimization sub-model are solved alternately.

2. The method for joint optimization based on user association and backhaul bandwidth configuration according to claim 1, characterized in that: The non-convex mixed integer fraction optimization model includes a continuous backhaul unit bandwidth configuration factor variable and a binary base station-user association variable.

3. The joint optimization method based on user association and backhaul bandwidth configuration according to claim 1, characterized in that: The alternating solution adopts an iterative updating optimization algorithm based on alternating optimization.

4. The method for joint optimization based on user association and backhaul bandwidth configuration according to claim 1, wherein: The backhaul unit bandwidth configuration optimization sub-model is based on fixed user association, and the dual-connectivity feasible user association optimization sub-model is based on a fixed backhaul bandwidth configuration factor.

5. The joint optimization method based on user association and backhaul bandwidth configuration according to claim 1, characterized in that: The non-convex mixed integer fraction optimization model maximizes the sum of user throughput utilities while considering user rate requirements by jointly optimizing user association variables and fronthaul bandwidth configuration variables.

6. The joint optimization method based on user association and backhaul bandwidth configuration according to claim 1, characterized in that: The backhaul unit bandwidth configuration optimization sub-model is a backhaul resource configuration optimization model under the condition of known user association, and the dual-connection feasible user association optimization sub-model is a user association optimization sub-model under the condition of known backhaul resource configuration factors.

7. The method for joint optimization based on user association and backhaul bandwidth configuration according to claim 1, characterized in that: The alternating solution includes obtaining an optimal backhaul bandwidth configuration mechanism by fixing user association variables; and solving a dual-connection feasible user association sub-sub-model by fixing backhaul bandwidth configuration factor variables.

8. The method for joint optimization based on user association and backhaul bandwidth configuration according to any one of claims 1 to 7, characterized in that: It also includes using a mathematical simulation method to verify the effectiveness and superiority of the joint optimization method based on user association and backhaul bandwidth configuration.

9. The method for joint optimization based on user association and backhaul bandwidth configuration according to claim 8, characterized in that: The non-convex mixed integer fractional optimization model has fractional terms in its optimization objectives and constraints, and has continuous variables and discrete variables. The optimization objectives and several constraints have a coupled product relationship between the continuous variables and the discrete variables.

10. The joint optimization method based on user association and backhaul bandwidth configuration according to claim 9, characterized in that: The method is based on a downlink transmission process of a two-layer heterogeneous network, wherein the network includes a macro base station and several openly accessible small base stations.

Images