User minimum spectrum efficiency enhancement method based on gradient descent

Abstract

The invention provides a user minimum spectrum efficiency enhancement method based on gradient descent, which is suitable for a non-orthogonal multiple access (Non-Orthogonal Multiple Access, NOMA) system of a large-scale multiple input multiple output (MIMO) without a cellular, and the method comprises the following steps of: firstly, selecting a user minimum spectrum efficiency (MME) of the user minimum spectrum efficiency enhancement method based on gradient descent, and selecting a user minimum spectrum efficiency (MME) of the user minimum spectrum efficiency enhancement method based on gradient descent, so that the user minimum spectrum efficiency enhancement method based on gradient descent is suitable for the non-cellular MIMO system. The method comprises the following three main steps of: firstly, calculating initial rates of all users through full power distribution to serve as initial conditions of subsequent optimization; secondly, dynamically optimizing a power distribution coefficient of the user by using a gradient descent method so as to maximize the minimum spectrum efficiency of the user in the system; and finally, taking an optimization result as an initial value of a continuous convex approximation (SCA) method, and improving the total spectrum efficiency of the system through step-by-step iteration until a convergence condition is met. According to the method, the gradient descent method and the SCA method are combined, so that the weak user performance and the fairness of system resource allocation are remarkably improved, and meanwhile, the total spectrum efficiency of the system is optimized. Compared with a traditional method with fixed assignment or without considering the minimum spectrum efficiency constraint, the method improves the calculation efficiency and adaptability, is suitable for dynamic and complex high-density user scenes, and provides an efficient and flexible solution for resource allocation of next-generation communication systems such as 5G, 6G and the like.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology, and its main application scenario is the non-cellular massive MIMO-NOMA system. Its main application is to solve the problem that the user's minimum spectrum efficiency constraint in the non-cellular massive MIMO-NOMA system is difficult to flexibly optimize. Specifically, it relates to a user minimum spectrum efficiency enhancement method based on gradient descent, which dynamically adjusts the user's minimum spectrum efficiency constraint by introducing a gradient optimization algorithm in the initial stage of iteration of the continuous convex approximation (SCA) method, replaces the traditional fixed assignment method, optimizes the initial conditions of power allocation, and improves weak user performance and system resource allocation fairness. The present invention significantly improves the overall spectrum efficiency of the system and accelerates the convergence of the algorithm through the combination of gradient descent optimization and SCA method. It is suitable for resource allocation and optimization in dynamic network environments, especially in high-density user access scenarios. It shows superiority and provides an efficient technical solution for multi-user access and spectrum efficiency improvement in next-generation communication systems such as 5G and 6G. Background Art

[0002] In the non-cellular massive multiple input multiple output (MIMO) non-orthogonal multiple access (NOMA) system, resource allocation optimization is one of the key links to improve system performance and user experience. Among them, power domain NOMA superimposes the data signals of multiple users in the power domain by utilizing the differences in user channel conditions, so that multiple users can share the same time-frequency resources, significantly improving the spectrum efficiency of the system. In this system, power allocation is one of the key links to maximize the overall system performance. However, in the process of resource allocation, how to balance the improvement of system performance and the optimization of weak user performance still faces many challenges.

[0003] In existing research, the goal of resource allocation is usually focused on maximizing the total spectrum efficiency of the system. To this end, the Successive Convex Approximation (SCA) is used to directly execute in the iterative process. The method transforms the non-convex optimization problem into multiple convex sub-problems and solves them step by step. However, these methods have shortcomings in the implementation of the minimum spectrum efficiency constraint: either a fixed minimum spectrum efficiency constraint is directly assigned, or the spectrum efficiency constraint of each user is not considered. The fixed assignment method lacks flexibility, which may lead to insufficient performance improvement for weak users and affect user fairness; ignoring the constraints on user spectrum efficiency may limit the overall performance of the system and make it difficult to meet the communication needs of different users in dynamic scenarios.

[0004] In addition, the optimization focus of existing methods is mostly on improving the overall spectrum efficiency, with less attention paid to the performance of weak users. Although the performance of weak users reflects the fairness of the system to a certain extent, its optimization is usually achieved indirectly and lacks a dedicated design. Especially in medium and high density user access scenarios, the problem of poor performance of weak users is particularly prominent, which directly affects the fairness and stability of system resource allocation.

[0005] Therefore, a more flexible optimization method is needed to effectively improve the performance of weak users while improving the overall spectrum efficiency of the system, dynamically adjust the minimum spectrum efficiency constraint to adapt to changes in user needs in different scenarios, and further improve the fairness of resource allocation and system performance. Summary of the invention

[0006] Inspired by the power allocation optimization strategy of the traditional SCA method, this paper proposes a method for enhancing the user's minimum spectrum efficiency based on gradient descent to address the problem of insufficient flexibility of the minimum spectrum efficiency constraint for each user in existing research. The proposed method introduces a gradient optimization algorithm in the initial stage of the iteration of the SCA method to dynamically adjust the user's minimum spectrum efficiency constraint, avoiding the limitations of the traditional fixed assignment method, effectively improving the performance of weak users while improving the overall spectrum efficiency of the system, and enhancing the fairness of resource allocation.

[0007] (I) Basic ideas and operations of the innovative points proposed by the present invention

[0008] The following specifically introduces the idea of ​​the method for enhancing the user minimum spectrum efficiency based on gradient descent proposed by the present invention. The core innovation of the present invention is to use the gradient descent method to dynamically optimize the user minimum spectrum efficiency to improve the fairness and total spectrum efficiency of the system, and further optimize the power allocation in combination with the SCA method. The design of this method includes the following main steps:

[0009] Step 1: Calculate the initial rates of all users through the full power allocation scheme to ensure that the initial spectrum efficiency of each user can be quickly obtained, thereby providing initial conditions for subsequent power allocation optimization.

[0010] Step 2: Based on the initial rate allocation, the power allocation coefficient is dynamically adjusted using the gradient descent method to maximize the minimum spectrum efficiency of users in the system. By combining the user's channel state information (CSI), the power allocation scheme is iteratively updated to optimize the performance of weak users and improve the fairness of resource allocation.

[0011] Step 3: Input the power allocation coefficient and spectrum efficiency optimized by the gradient descent method into the SCA method to further optimize the overall spectrum efficiency of the system.

[0012] 1) System Model

[0013] In the non-cellular massive MIMO-NOMA system, the total spectrum efficiency of the system is a key indicator to measure the system performance. Its optimization faces the trade-off between the complexity of power allocation and user fairness. At the same time, interference management has been preliminarily solved in the user clustering stage. In view of this non-convex optimization problem, this paper proposes a two-stage optimization method, which combines the gradient descent method with the SCA method in the classic literature to further improve the system performance. The total spectrum efficiency of the system can be expressed as:

[0014]

[0015] Wherein, formula (2) is the AP transmit power constraint; formula (3) is the SIC constraint.

[0016] In existing literature, a fixed lower limit of the minimum spectrum efficiency per user is often directly set as a constraint condition, but this fixed threshold is difficult to adapt to the dynamic changes of actual scenarios. In order to better improve user fairness, this paper designs a dynamic optimization scheme based on the gradient descent method to adjust the initial power allocation by maximizing the user's minimum spectrum efficiency. The specific optimization objectives are:

[0017]

[0018] The complete gradient descent-based user minimum spectral efficiency enhancement method is summarized in the following algorithm flow.

[0019] (II) Advantages of the present invention

[0020] The user minimum spectrum efficiency enhancement method based on gradient descent proposed in the present invention has the following advantages:

[0021] 1. The present invention dynamically optimizes the user's minimum spectrum efficiency through the gradient descent method, avoiding the limitation of the fixed minimum spectrum efficiency constraint in the traditional method, significantly improving the spectrum efficiency of weak users and enhancing the fairness of system resource allocation.

[0022] 2. By combining the gradient descent method with the SCA method, the present invention further improves the overall spectrum efficiency of the system through iterative solution on the basis of optimizing the minimum spectrum efficiency of users, thereby achieving a good balance between performance and fairness.

[0023] 3. The present invention adopts the gradient descent method to optimize the process of user minimum spectrum efficiency, avoiding the high computational complexity brought by global optimization or exhaustive methods, making system resource allocation more efficient and providing a low-complexity implementation solution for engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention provides the following drawings for illustration:

[0025] Attached Figure 1 Pseudo code for the user minimum spectrum efficiency enhancement method based on gradient descent DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the objectives, technical solutions and advantages of the present invention, a complete description is given below in conjunction with specific examples and drawings.

[0027] Attached Figure 1 The pseudo code of the method for enhancing the user minimum spectrum efficiency based on gradient descent is as follows:

[0028] Step 1: Initialize the power allocation coefficient Learning rate b, maximum number of iterations L, using the full power allocation scheme to calculate the initial frequency efficiency SE of each user nk .

[0029] Step 2: According to the initial power allocation scheme, find the minimum spectrum efficiency minSE nk It is used as the target initial condition for subsequent iterative optimization.

[0030] Step 3: In each iteration, use the gradient of spectral efficiency Update power allocation coefficient γ mnk ,Right now

[0031] At the same time, ensure that the power allocation coefficient satisfies the non-negative constraint γ mnk ≥0.

[0032] Step 4: Recalculate the spectrum efficiency of all users using the new power allocation coefficient and update the minimum spectrum efficiency minSE nk .

[0033] Step 5: Determine the objective function minSE nk Whether the convergence condition is met or the number of iterations reaches the maximum value L, that is, if the convergence condition is met or the maximum number of iterations is reached, stop the iteration and go to step 6; otherwise, return to step 3 and continue iterative optimization.

[0034] Step 6: Output the final spectrum efficiency SE of all users nk And the optimized power allocation coefficient γ mnk. .

Claims

1. A method for enhancing user minimum spectrum efficiency based on gradient descent, characterized in that: This method optimizes the power allocation coefficient of users by gradient descent to maximize the minimum spectral efficiency of users in the cell-free massive MIMO-NOMA system. The main process of this method includes: In the first step, the initial spectrum efficiency of all users is calculated using the full power allocation scheme, which serves as the initial condition for subsequent gradient optimization. In the second step, the power allocation coefficient is dynamically adjusted based on the gradient descent method. The update formula is: Here ▽SE nk is the gradient of spectral efficiency, γ mnk is the power allocation coefficient, b is the learning rate; The third step is to iteratively update the power allocation coefficient and the user's minimum spectrum efficiency until the convergence condition is met or the maximum number of iterations is reached, and then output the final optimization result.

2. The method for enhancing user minimum spectrum efficiency based on gradient descent according to claim 1, characterized in that: When the gradient descent method dynamically adjusts the power allocation coefficient, ensure that the power allocation coefficient satisfies the non-negativity constraint 3. The method for enhancing user minimum spectrum efficiency based on gradient descent according to claim 1, characterized in that: This method updates the user's minimum spectral efficiency minSE in real time in each iteration. nk , and use the learning rate b to dynamically adjust the step size of gradient descent to improve the convergence speed and stability of the algorithm.

4. The method for enhancing user minimum spectrum efficiency based on gradient descent according to claim 1, characterized in that: By optimizing the user's minimum spectrum efficiency, the proposed method can significantly improve the performance of weak users and enhance the fairness of system resource allocation.

5. The method for enhancing user minimum spectrum efficiency based on gradient descent according to claim 1, characterized in that: This method is suitable for dynamic and complex network scenarios. By dynamically adjusting the power allocation scheme, the overall spectrum efficiency and system robustness of the non-cellular massive MIMO-NOMA system are improved.

Images