CN111277308A - Machine Learning-Based Wavewidth Control Method - Google Patents

Abstract

Aiming at the problem that user coverage and spectral efficiency are improved through intelligent beam width control in a millimeter wave (mmWave) -non-orthogonal multiple access (NOMA) network, an intelligent beam width control mechanism based on Machine Learning (ML) is provided, namely angle domain information of users at several moments in the future is predicted through Gaussian Process Machine Learning (GPML), user grouping is conducted through an unsupervised learning method according to the predicted angle domain information, accurate channel state information is obtained through methods such as beam tracking and channel estimation, and finally the optimal beam width is obtained through a Deep Neural Network (DNN) according to the channel state information, the angle domain information and the user grouping information. The beam tracking efficiency can be obviously improved through angle domain information prediction; by training DNN offline, the time for solving online can be saved; the channel state information, the angle domain information and the user grouping information of the users are directly mapped to the optimal beam width through the DNN, so that the user coverage and the spectral efficiency of the system can be remarkably improved.

Description

Machine Learning-Based Wavewidth Control Method

technical field

The present invention relates to the realization of flexible beam width control according to the user's angular domain information in multi-user scenarios, to be precise, using the angular domain information (ADI) to realize millimeter wave (mmWave)-non-orthogonal by means of machine learning (ML) Efficient transmission and flexible coverage of multiple access (NOMA) systems belong to the technical field of wireless communication.

Background technique

Due to the abundant spectrum resources in the high frequency band, the millimeter wave (mmWave) frequency band has become the key technology of the fifth-generation mobile communication system. However, the implementation of millimeter wave communication still faces serious path loss. Therefore, in order to compensate the path loss of millimeter-wave communication, the millimeter-wave communication system usually configures massive multiple-input multiple-output (m-MIMO) technology at the transceiver end to achieve directional transmission. The use of space division multiplexing can significantly improve the energy of the system. efficiency and spectral efficiency. Considering the distribution of users in three-dimensional space, compared with horizontal two-dimensional beamforming, three-dimensional beamforming can achieve more flexible coverage by adjusting the horizontal and vertical angles, thereby further improving system performance.

Due to the large-scale antenna array of mmWave transceivers, a purely digital transceiver architecture will consume huge energy and hardware resources. To balance hardware, energy cost and system performance, a hybrid transceiver architecture is a more practical choice. However, non-orthogonal multiple access (NOMA) technology is integrated into cellular networks when the number of users is greater than the number of RFs due to its limited number of radio frequency links (RFs) and to reduce the energy consumption of the base stations. Different users can significantly improve the user coverage and spectrum efficiency of the system by reusing time-frequency space resources.

The traditional two-dimensional (2D) bandwidth control is based on a linear antenna array. The width of the beam is controlled by adjusting the number of activated antenna elements. The more the number of activated antenna elements, the narrower the beam and the more concentrated the energy. However, in the mmWave-NOMA scenario, the narrow beam will reduce the coverage of the beam, that is, reduce the user coverage capability of the system, and then damage the spectral efficiency of the system. The three-dimensional (3D) wave width control proposed by the present invention is to control the beam width in both horizontal and vertical directions by adjusting the number of antenna elements of the planar antenna array in the two dimensions of x-y. The user's quality of service and the maximization of system performance cannot be guaranteed by the empirical method of 3dB bandwidth. Although the optimal system performance can be achieved through the exhaustive search method, the complexity of the system will increase exponentially with the increase of the antenna size, the number of RF and the number of users. In m-MIMO and dense user-dense cellular scenarios, An exhaustive search approach is also impractical.

At present, machine learning is in the ascendant, and it has been widely used in the fields of pattern recognition and artificial intelligence to the design and optimization of communication systems. Gaussian Process Machine Learning (GPML), Unsupervised Learning and Deep Learning are important branches of Machine Learning. Among them, GPML is developed from statistical learning theory and Bayesian theory. Compared with methods such as neural networks and support vector machines, GMPL has the advantages of easy implementation, adaptive acquisition of hyperparameters, flexible nonparametric reasoning and probabilistic meaningful output. It is widely used in the field of time series forecasting. Unsupervised learning, also known as clustering, compared with supervised learning, unsupervised learning does not require labels, and is mainly used to mine data patterns, structures or laws. Due to its powerful data fitting ability, deep learning has been widely used in the design and optimization of communication systems, especially in dealing with communication problems that are difficult to express with mathematical models. Based on a large amount of historical data, deep learning is used to establish the mapping relationship between system features and system parameters, thereby improving system performance.

In addition, communication systems generally have high requirements for real-time performance. Simply calculating the corresponding parameters of the system (such as beam widths, etc.) through algorithms may not meet the requirements of low latency due to the complexity of the calculation process. Therefore, the machine learning (ML) method can be introduced again, and the deep neural network (DNN) can be trained by using the existing calculated data to obtain "experience", and then the input of the new state variables can also be more accurate. system parameters, the system performance has been significantly improved.

SUMMARY OF THE INVENTION

The present invention considers millimeter wave (mmWave)-non-orthogonal multiple access (NOMA) scenarios and makes predictions based on multi-user historical angle domain information (ADI), and proposes an intelligent wave width control method based on machine learning (ML), that is, through Gaussian Process machine learning (GPML) to predict the ADI of users in the future, and then use the unsupervised learning method to group users according to the predicted ADI, and then obtain the ADI and channel gain information through beam tracking and channel estimation, and finally according to the user grouping information, ADI and channel gain information through the deep neural network to obtain the optimal beam width that can maximize the spectral efficiency of the system. By training the DNN offline, the real-time performance of the system can be improved.

Description of drawings

Fig. 1 is a flow chart of the realization of the bandwidth control.

Fig. 2 is a structural diagram of a deep neural network for wave width control.

FIG. 3 is a system model diagram of a single base station multi-user scenario.

Figure 4 is the system SE comparison curve of the optimal BC scheme without bandwidth control, BC based on 3dB bandwidth, BC based on machine learning and exhaustive search.

Figure 5 is a comparison curve of user coverage capability of no BC, BC based on 3dB bandwidth, BC based on machine learning, and the optimal BC scheme based on exhaustive search. The ordinate is the proportion of users who meet the service quality.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

Referring to FIG. 1 , the flow chart of the realization of the bandwidth control, based on the offline database (including angle domain information (ADI), channel state information, user grouping information), the optimal beamwidth is obtained by offline exhaustive search. With the optimal beam width as the label and the corresponding offline data information as the feature, the deep neural network (DNN) is trained to obtain the DNN model. In the online implementation process, the real-time angle domain information (ADI), channel state information, user grouping information and other features are input into the DNN to obtain near-optimal beam widths, thereby improving system performance.

Referring to Fig. 2, the neural network structure diagram for realizing the bandwidth control takes the optimal beamwidth as the label, and takes the ADI, channel gain and grouping information as the input features, and the input features obtain the corresponding beamwidth output through the DNN. The optimal beamwidth) is compared with the calculation loss, and the neural network parameters are continuously updated through gradient descent and reverse transfer until the difference between the output of the DNN and the label is lower than a certain threshold.

Referring to Figure 3, the system model diagram of the single base station multi-user scenario, the millimeter wave base station is configured with a hybrid architecture transmitter. Each user predicts the future ADI according to the historical ADI, and feeds back the prediction result to the base station. The base station groups NOMA users according to the predicted ADI, and then performs beam scanning in each NOMA group to obtain accurate ADI, thereby completing link establishment.

Referring to Fig. 4, the average spectral efficiency of 1000 time slots is plotted as a function of signal-to-noise ratio (SNR). The average spectral efficiency increases with increasing SNR. The bandwidth control method based on machine learning is close to the optimal solution obtained by exhaustive search, and is significantly better than the bandwidth control method based on 3dB bandwidth and the traditional NOMA transmission method without bandwidth control. The bandwidth control method can significantly improve the spectral efficiency of the system.

Referring to Fig. 5, the average proportion of users who satisfy the quality of service (QoS) in 1000 time slots accounts for the total number of users. As the signal-to-noise ratio (SNR) increases, the number of users satisfying QoS gradually increases. The bandwidth control method based on machine learning is close to the optimal solution obtained by exhaustive search, and is significantly better than the bandwidth control method based on 3dB bandwidth and the traditional NOMA transmission method without bandwidth control. The bandwidth control method can significantly improve the user coverage capability of the system.

In summary, we can obtain the beamwidth with better performance of each time slot by the ML method, thus verifying the concept of the machine learning-based beamwidth control.

The above is only an example of the present invention, and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection of the present invention. within the range.

Claims (1)

1. An intelligent wave width control method based on Machine Learning (ML), wherein a plurality of users simulating millimeter wave architecture exist in a millimeter wave network, a base station mixing millimeter wave architecture, millimeter wave transmission is carried out between the base station and the users, the base station adjusts the wave beam width of each cluster through the machine learning method to improve the spectrum efficiency and the user coverage capability, and the specific method comprises the following steps:

(1) predicting angle domain information based on a Gaussian process machine learning method: the historical angle domain information is used as input, the angle domain information of a plurality of time slots in the future is predicted, and beam tracking and user grouping of millimeter wave communication are further assisted;

(2) unsupervised learning based user grouping: according to the predicted angle domain information, user clustering is carried out by using an unsupervised learning method, and the base station carries out non-orthogonal multiple access transmission on the users of each cluster;

(3) wave width control based on a deep neural network: obtaining channel gain information through channel estimation, taking an angle domain, channel gain and user clustering information as the input of a deep neural network, taking the optimal beam width of each cluster obtained through exhaustive search as a corresponding label, training the deep neural network offline, and establishing a mapping relation between system parameters and the optimal beam width;

(4) the online realization comprises the following steps: and transferring the offline-trained wave width control neural network model to an actual millimeter wave system, and correcting the model by using a transfer learning method when the network topology in the system is changed violently.

Images