A resource allocation method for non-orthogonal multiple access networks assisted by reconfigurable intelligent surfaces

By adopting a non-orthogonal multi-access network resource allocation method with reconstructible intelligent surface assist in wireless networks, and using deep learning algorithms to optimize resource allocation, the problem that traditional multi-access access solutions cannot improve spectrum efficiency is solved, and dynamic Efficient user resource allocation in the environment.

CN116600396BActive Publication Date: 2024-11-08TIANTANSOFT
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202310706782.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-15
Publication Date
2024-11-08
Estimated Expiration
2043-06-15

AI Technical Summary

Technical Problem

Traditional multiple access solutions cannot effectively improve the spectrum efficiency and user connection quality of wireless networks, especially in 5G and later networks where data and access volumes are rapidly increasing.

Method used

Using the reconstructible intelligent surface-assisted non-orthogonal multiple access network resource allocation method, the channel model is established, and the user mobility and clustering are predicted using long and short-term memory algorithms, K-mean Gaussian hybrid model algorithms and deep Q network-based algorithms. , jointly determine the phase shift matrix and power allocation strategy, and optimize the resource allocation strategy to maximize communication rate.

Benefits of technology

User resource allocation in a multi-input single-output non-orthogonal multi-access network is realized in a dynamic environment, which improves spectrum efficiency and user connection quality, and has good application value.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116600396B_ABST
    Figure CN116600396B_ABST
Patent Text Reader

Abstract

The present invention discloses a non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface. First, a channel model between a user and a base station is established. Then, a long short-term memory algorithm, a Gaussian mixture model algorithm based on K-means, and an algorithm based on a deep Q network are used to respectively predict user mobility, user clustering, and jointly determine a phase shift matrix and a power allocation strategy. Finally, the optimal resource allocation strategy under the network is optimized. The present invention provides a novel multi-input single-output non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface in a dynamic environment, which has good application value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of wireless communications, and in particular to a non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface. Background Art

[0002] As the demand for large capacity in wireless networks increases, traditional multiple access schemes cannot guarantee connection quality. Therefore, the pursuit of spectrum efficiency has become a major concern for wireless networks, especially in 5G and beyond networks where the amount of data and access are exploding. Although various technologies have been called to improve spectrum efficiency, such as massive multiple-input multiple-output and millimeter-wave communications, ultra-large-scale and ubiquitous wireless connectivity, they are still far from improving spectrum and energy efficiency.

[0003] Non-orthogonal multiple access is a novel access technology that uses the non-orthogonality principle to enable multiple users to share time domain, frequency domain or code domain resources. The most attractive is the power domain non-orthogonal multiple access technology, the core idea of ​​which is to superimpose the signals of two users with different power and utilize the spectrum more efficiently by opportunistically exploring different channel conditions of users. Compared with the orthogonal multiple access scheme that only allocates a single radio resource to one user, the non-orthogonal multiple access technology applies superposition coding and continuous interference cancellation techniques at the base station and user end respectively. In addition, in order to further improve the spectrum efficiency and user connectivity of non-orthogonal multiple access wireless networks, reconfigurable smart surfaces have great advantages. It can reconstruct the wireless propagation environment by integrating a large number of low-cost passive reflective elements on a plane, thereby significantly improving the performance of wireless communication networks. The elements on the reconfigurable smart surface can independently reflect the received incident signal by controlling its amplitude and / or phase, so that the received signal from the base station to the user can be enhanced. Compared with some communication assistance technologies (such as amplify-and-forward and decode-and-forward relay), wireless networks assisted by reconfigurable smart surfaces consume less energy. Therefore, inspired by the advantages of reconfigurable smart surfaces, wireless networks assisted by reconfigurable smart surfaces have been regarded as one of the candidate solutions for the next generation of wireless communication systems. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention aims to propose a non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface.

[0005] In order to achieve the above purpose, the present invention adopts the following technical methods:

[0006] A non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface comprises the following steps:

[0007] Step 1: Establish a channel model between the communication user and the base station;

[0008] The channel model includes an N-antenna base station, L mobile users, and a reconfigurable smart surface with K reflective elements; the direct communication link between the N-antenna base station and the mobile users is blocked, and the communication link can only be established through the reconfigurable reflective surface, and the L mobile users will be divided into M classes;

[0009] The channels from the N-antenna base station to the reconfigurable smart surface and from the reconfigurable smart surface to the l-th mobile user in the M-th class are defined as G∈C K×N and H represents conjugate; in addition, for the reconfigurable smart surface, its reflection coefficient matrix can be expressed as where β k and θ k denote the amplitude and phase of the kth element respectively, then the signal received by the lth mobile user in the mth class is:

[0010]

[0011] where n m,l ~CN(0,δ 2 ) indicates zero mean and variance δ 2 Additive Gaussian white noise, ω m and x m is the beamforming and transmission symbol of the mth class; assuming that the lth i and l j Mobile users, whose decoding order satisfies Then the first i and l j The signal to interference noise ratio of each user is expressed as:

[0012]

[0013]

[0014] in, and Respectively represent the remaining users in the mth class, The information symbol of the mobile user and the lth i The information symbol of each mobile user; then, the corresponding received communication rate can be calculated by Shannon's formula R = log2(1+τ), where τ represents the signal-to-noise ratio; the optimization of the resource allocation strategy is expressed as maximizing the number of all users at T = {t0, t1, …, t n}Total communication rate during the time period;

[0015] Step 2: Use the long short-term memory algorithm, the K-means Gaussian mixture model algorithm, and the deep Q-network-based algorithm to predict the mobility of mobile users, mobile user clustering, and jointly determine the phase shift matrix and power allocation strategy;

[0016] Using the acceptance-rejection sampling method, a set of initial positions are randomly generated for L mobile users. According to the long short-term memory algorithm, we can predict the time from t1 to t n The position at each moment can be expressed as At each moment, the Gaussian mixture model algorithm based on K means is called for clustering; based on the mth class, the model coefficients in the Gaussian mixture model algorithm of K means are and It can be expressed as:

[0017]

[0018]

[0019]

[0020] Among them, h l , and μ m They represent the channel of the lth mobile user, the expected value of the evaluation coefficient of the Gaussian distribution corresponding to the position of the lth user in the mth class, and the mean of the Gaussian distribution corresponding to the mth class respectively; finally, according to the clustering results, the N-antenna base station is set as the agent, and the state space, action space and reward are defined as the current phase, the adjustable phase and the difference of the total communication rate of all users at adjacent times;

[0021] Step 3: Based on step 2, optimize the optimal resource allocation strategy under the deep Q network; when the neural network reaches convergence after training, the optimal resource allocation strategy can be output.

[0022] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0023] The present invention also provides a computer device, comprising a processor and a memory, wherein the memory is used to store a computer program; and the processor is used to execute the computer program to implement the above method.

[0024] The beneficial effect of the present invention is that: by using the present invention, resources can be allocated for users in a multi-input single-output non-orthogonal multiple access network assisted by a reconfigurable intelligent surface in a dynamic environment, and the present invention has good application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Schematic diagram of the overall idea of ​​the method in the embodiment of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with the accompanying drawings. It should be noted that this embodiment is based on the technical method and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to this embodiment.

[0027] This embodiment provides a method for allocating resources in a non-orthogonal multiple access network assisted by a reconfigurable intelligent surface. By this method, resource allocation of a multi-input single-output non-orthogonal multiple access network assisted by a reconfigurable intelligent surface in a dynamic environment is achieved. Figure 1 As shown, the method of this embodiment first establishes a channel model between the communication user and the base station, and then uses the long short-term memory algorithm, the Gaussian mixture model algorithm based on K-means, and the algorithm based on the deep Q network to respectively predict the user's mobility, user clustering, and jointly determine the phase shift matrix and power allocation strategy. Finally, based on step 2, the optimal resource allocation strategy under the network is optimized. When the neural network reaches convergence after training, the optimal resource allocation strategy can be output.

[0028] The reconfigurable intelligent surface-assisted non-orthogonal multiple access network resource allocation method specifically comprises the following steps:

[0029] Step 1: Establish a channel model between the communication user and the base station.

[0030] The channel model includes an N-antenna base station, L mobile users, and a reconfigurable smart surface with K reflective elements. The direct communication link between the N-antenna base station and the mobile users is blocked, and the communication link can only be established through the reconfigurable reflective surface, and the L mobile users will be divided into M classes.

[0031] The channels from the N-antenna base station to the reconfigurable smart surface and from the reconfigurable smart surface to the l-th mobile user in the m-th class are defined as G∈C K×N and H represents conjugate; in addition, for the reconfigurable smart surface, its reflection coefficient matrix can be expressed as where β k and θ k denote the amplitude and phase of the kth element respectively, then the signal received by the lth mobile user in the mth class is:

[0032]

[0033] where n m,l ~CN(0,δ 2 ) indicates zero mean and variance δ 2Additive Gaussian white noise, ω m and x m is the beamforming and transmission symbol of the mth class. Assume that the lth i and l j Mobile users, whose decoding order satisfies Then the first i and l j The signal to interference noise ratio of each user is expressed as:

[0034]

[0035]

[0036] in, and Respectively represent the remaining users in the mth class, The information symbol of the mobile user and the lth i The corresponding received communication rate can be calculated by Shannon's formula R = log2(1+τ), where τ represents the signal-to-noise ratio; the optimization of resource allocation strategy is expressed as maximizing the number of all users in T = {t0, t1, ..., t n}The total communication rate during the time period.

[0037] Step 2: Use the long short-term memory algorithm, the K-means Gaussian mixture model algorithm, and the deep Q-network-based algorithm to predict the mobility of mobile users, mobile user clustering, and jointly determine the phase shift matrix and power allocation strategy.

[0038] Without loss of generality, we use the acceptance-rejection sampling method proposed by von Neumann to randomly generate a set of initial positions for L mobile users. According to the long short-term memory algorithm, we can predict the time from t1 to t n The position at each moment can be expressed as At each moment, the Gaussian mixture model algorithm based on K means is called for clustering. Based on the mth class, the model coefficients in the Gaussian mixture model algorithm based on K means are and It can be expressed as:

[0039]

[0040]

[0041]

[0042] Among them, h l , and μ mThey represent the channel of the lth mobile user, the expected value of the evaluation coefficient of the Gaussian distribution corresponding to the position of the lth user in the mth class, and the mean of the Gaussian distribution corresponding to the mth class. Finally, according to the clustering results, the N-antenna base station is set as the agent, and the state space, action space and reward are defined as the current phase, the adjustable phase and the difference in the total communication rate of all users at adjacent times.

[0043] Step 3: Based on step 2, optimize the optimal resource allocation strategy under the deep Q network. When the neural network reaches convergence after training, the optimal resource allocation strategy can be output.

[0044] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical methods and concepts, and all of these changes and modifications should be included in the protection scope of the claims of the present invention.

Claims

1. A non-orthogonal multiple access network resource allocation method assisted by a reconfigurable intelligent surface, characterized in that: The steps include: Step 1: Establish a channel model between the communication user and the base station; The channel model includes an N-antenna base station, L mobile users, and a reconfigurable smart surface with K reflective elements; the direct communication link between the N-antenna base station and the mobile users is blocked, and the communication link can only be established through the reconfigurable reflective surface, and the L mobile users will be divided into M classes; The channels from the N-antenna base station to the reconfigurable smart surface and from the reconfigurable smart surface to the l-th mobile user in the m-th class are defined as G∈C K×N and H represents conjugate; in addition, for the reconfigurable smart surface, its reflection coefficient matrix can be expressed as where β k and θ k denote the amplitude and phase of the kth element respectively, then the signal received by the lth mobile user in the mth class is: where n m,l denotes zero mean and variance δ 2 Additive Gaussian white noise, ω m and x m is the beamforming and transmission symbol of the mth class; assuming that the lth i and l j Mobile users, whose decoding order satisfies Then the first i and l j The signal to interference noise ratio of each user is expressed as: in, and Respectively represent the remaining users in the mth class, The information symbol of the mobile user and the lth i The information symbol of each mobile user; then, the corresponding received communication rate can be calculated by Shannon's formula R = log2(1+τ), where τ represents the signal-to-noise ratio; the optimization of the resource allocation strategy is expressed as maximizing the number of all users at T = {t0, t1, …, t n }Total communication rate during the time period; Step 2: Use the long short-term memory algorithm, the K-means Gaussian mixture model algorithm, and the deep Q-network-based algorithm to predict the mobility of mobile users, mobile user clustering, and jointly determine the phase shift matrix and power allocation strategy; Using the acceptance-rejection sampling method, a set of initial positions are randomly generated for L mobile users. According to the long short-term memory algorithm, t1 to t n The position at each moment can be expressed as At each moment, the Gaussian mixture model algorithm based on K means is called for clustering; based on the mth class, the model coefficients in the Gaussian mixture model algorithm of K means are and It can be expressed as: Among them, h l , and μ m They represent the channel of the lth mobile user, the expected value of the evaluation coefficient of the Gaussian distribution corresponding to the position of the lth user in the mth class, and the mean of the Gaussian distribution corresponding to the mth class respectively; finally, according to the clustering results, the N-antenna base station is set as the agent, and the state space, action space and reward are defined as the current phase, the adjustable phase and the difference of the total communication rate of all users at adjacent times; Step 3: Based on step 2, optimize the optimal resource allocation strategy under the deep Q network; when the neural network reaches convergence after training, the optimal resource allocation strategy can be output.

2. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to claim 1 is implemented.

3. A computer device, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program; when the processor is used to execute the computer program, the method described in claim 1 is implemented.

Citation Information

Patent Citations

  • Reflecting surface assisted non-orthogonal multiple access communication system design method

    CN110225538A

  • Intelligent reflector-assisted SM-NOMA system resource allocation method

    CN112865893A

  • Communication resource allocation method and apparatus, and electronic device

    CN113163501A

  • Intelligent reflecting surface assisted non-orthogonal multiple access system power distribution method and system

    CN115843034A

  • Intelligent reflecting surface assisted UAV-NOMA system resource allocation method

    CN116156429A