Communication interference suppression method and apparatus

Directly processing signals through neural networks to suppress wireless communication interference, solving the calculation problem of interference from multiple interference sources in multiple inlet and exit scenarios, improving the efficiency and accuracy of wireless communication, and is suitable for complex scenarios and new generation network architectures.

WO2024250148A1PCT designated stage expired Publication Date: 2024-12-12HUAWEI TECH CO LTD

Patent Information

Application Number
PCT/CN2023/098330
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-06-05
Publication Date
2024-12-12

AI Technical Summary

Technical Problem

In the multi-in and multi-out scenario, it is difficult for the prior art to effectively calculate and suppress signal interference caused by multiple interference sources, resulting in low wireless communication efficiency. Especially when the number of interference sources is large and the real-time changes, the efficiency and accuracy of the traditional methods are insufficient. .

Method used

The neural network is used to suppress the received signal wireless communication interference. The trained neural network directly processes the signal to reduce interference, without calculating the interference source or performing interference distribution calculations. It is suitable for multi-input and multiple outflow and complex scenarios, improving interference Accuracy and efficiency of inhibition.

Benefits of technology

It realizes efficient and accurate interference suppression in multiple inlet and exit scenarios, reduces system complexity and error, improves the efficiency and signal-to-noise ratio of wireless communication, and is suitable for 5G and 6G network architectures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023098330_12122024_PF_FP_ABST
    Figure CN2023098330_12122024_PF_FP_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of wireless communications, and particularly to the technical field of signal interference suppression, and provides a communication interference suppression method and an apparatus. The communication interference suppression method of the present application is applied to a first communication main body, and comprises: acquiring a signal from a second communication main body, and using a neural network to perform wireless communication interference suppression on the signal, so as to obtain an interference-suppressed signal. In the present application, the neural network is used to perform wireless communication interference suppression on signals. Signals can be inputted into the neural network so as to perform interference suppression on the signals without having to perform interference distribution calculation, thus achieving higher interference suppression efficiency and accuracy and improving wireless communication efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Communication interference suppression method and device Technical Field

[0001] The embodiments of the present invention relate to the field of wireless communication technologies, in particular to the field of signal interference suppression technologies, and specifically to a communication interference suppression method, device, electronic device, storage medium, and program product. Background Art

[0002] Wireless communication technology involves transmitting signals between multiple communicating entities without using conductors or cables. During wireless communication, signals transmitted between any of these entities may be interfered with by other entities. This poses a technical challenge in accurately and efficiently mitigating interference in wireless communications.

[0003] In the prior art, communication subject A sends a signal S to communication subject B, and communication subject B receives a signal S' that has been interfered with by an interference source. The interference distribution can be calculated based on the signal S' and relevant information about the interference source, and the interference distribution can be used to perform corresponding interference suppression on the signal S', so that communication subject B can obtain a more accurate signal.

[0004] The above-mentioned existing technology is applicable to situations where there are fewer interference sources. With the development of wireless communication technology, for example, in scenarios such as multiple-input and multiple-output, the number of interference sources is large, and it is difficult to accurately calculate the interference distribution. This makes the efficiency of interference suppression using the above-mentioned existing technology low, thereby affecting the efficiency of wireless communication.

[0005] Summary of the Invention

[0006] The present application provides a communication interference suppression method and device, which can accurately and efficiently suppress communication interference, and is particularly suitable for communication interference suppression in complex scenarios.

[0007] In a first aspect, the present application relates to a communication interference suppression method, applied to a first communication subject, comprising: obtaining a signal from a second communication subject, and performing wireless communication interference suppression on the signal using a neural network to obtain an interference suppression signal.

[0008] The signal obtained by the first communication subject from the second communication subject is interfered with. The wireless communication interference may come from some interference sources, or may be loss during signal transmission (loss may include noise, for example).

[0009] Of course, in some cases, the signal received by the first communication subject from the second communication subject is not interfered with, and the signal can still be processed by the communication interference suppression method of the embodiment of the present application.

[0010] In the embodiment of the present application, a neural network is used to suppress wireless communication interference of the signal. Only the received signal is used as input, and it is not concerned about what interference source the signal is interfered with. Therefore, there is no need to solve each interference source through calculation, nor is there any need to calculate the interference distribution of the signal. The signal only needs to be input into the neural network to suppress the interference of the signal. This can greatly reduce the complexity of communication interference suppression and has higher interference suppression efficiency. It is particularly suitable for complex scenarios such as multiple inputs and multiple outputs and multiple interference sources.

[0011] On the one hand, in a multi-input and multi-output scenario, especially in a scenario where there are many interference sources and the interference changes in real time, the communication interference suppression method of the embodiment of the present application can still use the signal at each moment as input regardless of how the interference changes in real time, without paying attention to the interference source, that is, there is no need to solve each interference source. In a scenario where there are a large number of interference sources, the error in solving the interference source can be greatly reduced, and there is no need to perform interference distribution calculation in real time, which can greatly reduce the error in calculating the interference distribution in real time, thereby improving the accuracy and efficiency of interference suppression. In addition, this process will not interfere with the normal operation of the communication system, and therefore will not affect the performance of the communication system. For example, the communication interference suppression method of the embodiment of the present application will not interfere with the normal operation of the communication system based on multi-input and multi-output technology, and therefore will not reduce the performance gain of the communication system.

[0012] On the other hand, the embodiments of the present application do not require real-time and frequent acquisition of neighboring cell scheduling information through signaling, do not need to know the signal type of user equipment in the neighboring cell, and do not incur signaling overhead for knowing the signal type of user equipment in the neighboring cell. As a result, the communication interference suppression method of the embodiments of the present application has higher interference suppression efficiency.

[0013] On the other hand, the trained neural network can ensure a certain degree of interference suppression accuracy. Therefore, the interference suppression signal obtained by using the neural network to suppress wireless communication interference is closer to the original signal sent by the transmitter.

[0014] Higher accuracy can improve wireless communication efficiency.

[0015] On the other hand, taking wireless communication between cells of base stations in a cellular system as an example, the embodiments of the present application only focus on the received signals. In the process of suppressing wireless communication interference, the spectrum resources of each cell will not be affected, and therefore the spectrum efficiency of the network will not be reduced.

[0016] In summary, the embodiments of the present application, by inputting signals into a neural network to obtain interference suppression signals, can adapt to large-scale multiple-input multiple-output technologies, cell-free multiple-input multiple-output technologies, etc. Since 5G and 6G technologies mostly adopt large-scale multiple-input multiple-output technologies and cell-free multiple-input multiple-output technologies, the embodiments of the present application can adapt to network architecture deployments such as 5G and 6G.

[0017] In addition, the embodiments of the present application have better performance in various aspects such as communication interference estimation accuracy, signal-to-noise ratio, and throughput.

[0018] In a possible embodiment, before acquiring the signal from the second communication subject, the method further includes: sending a training mode start instruction to start neural network training.

[0019] The training mode start instruction is used to instruct the interference source, the first communication subject, and the second communication subject to start wireless communication in a non-interference mode.

[0020] In the non-interference mode, the second communication subject obtains the first signal that is not interfered by the interference source.

[0021] For example, the first communication subject may send a training mode start instruction to the second communication subject and the interference source to start neural network training. After receiving the training mode start instruction, the second communication subject may send a non-interference signal that is not interfered with by the interference source to the first communication subject.

[0022] Exemplarily, when the interference source receives a training mode start-up instruction, the interference source can, for example, send signal S1, and the second communication subject can, for example, send signal S2. Signal S1 can, for example, be orthogonal to signal S2 or in a different frequency band. This can ensure that the signal S1 of the interference source does not interfere with the signal S2 sent by the second communication subject, and the second communication subject can obtain the first signal that is not interfered with by the interference source.

[0023] Exemplarily, the training mode start instruction may include indication data and information data. The indication data may be used to indicate the start of the non-interference mode between the first communication subject, the second communication subject and the interference source. The information data may be used to indicate the working time, working duration and other information of the non-interference mode.

[0024] In a possible embodiment, an initial neural network is trained using non-interference signal samples and interfered signal samples to obtain a neural network.

[0025] The interference-free signal sample is obtained based on the first signal, and the interfered signal sample is obtained based on the second signal, where the second signal is a signal obtained by the second communication subject and interfered with by an interference source.

[0026] In the embodiment of the present application, the difference between the interference-free signal samples and the interfered signal samples can characterize the interference signal, and the interference signal can be used to suppress communication interference. Therefore, by training the initial neural network using the interference-free signal samples and the interfered signal samples, the obtained neural network can be used to accurately estimate the interference signal and suppress communication interference.

[0027] In one possible embodiment, the neural network may be trained using, for example, the following operations: using an initial neural network to process interference-impaired signal samples to obtain sample interference-suppressed signals; determining feedback values ​​based on the sample interference-suppressed signals and interference-free signal samples; and adjusting parameters of the initial neural network based on the feedback values ​​to obtain the neural network.

[0028] The initial neural network can be understood as a neural network with initial network parameters. The process of training the initial neural network can adjust the network parameters. The network parameters of the neural network obtained after training make the performance of the neural network for wireless communication interference suppression better.

[0029] Since the sample interference suppression signal can be understood as the result obtained after interference suppression of a neural network with initial network parameters, the result obtained after interference suppression under ideal conditions should be an undisturbed signal. Therefore, the embodiment of the present application uses the undisturbed interference-free signal sample as a label. Based on the sample interference suppression signal and the interference-free signal sample, the feedback value determined can characterize the interference suppression performance of the current neural network.

[0030] Since the feedback value can characterize the interference suppression performance of the current neural network, by adjusting the parameters of the initial neural network according to the feedback value, a neural network with better interference suppression performance can be obtained, thus completing the neural network training.

[0031] Exemplarily, the initial neural network may be a pre-trained neural network.

[0032] A pre-trained neural network can be understood as a neural network obtained through prior training on a large-scale data set. The goal of pre-training is to train a general neural network that can be fine-tuned or transferred using a smaller business data set in subsequent tasks.

[0033] The embodiments of the present application train a neural network and utilize the trained neural network to suppress communication interference. The communication interference suppression method is executed on a first communication subject, such as a user device or a base station. Small-scale neural network training is supported, and therefore, higher requirements are placed on the training speed and efficiency of the neural network. A pre-trained neural network that has been pre-trained has fixed parameters and only requires fine-tuning of the parameters. Therefore, training the pre-trained neural network is faster and more efficient.

[0034] In one possible embodiment, for example, the following embodiments can be used to implement a specific example of processing interfered signal samples using an initial neural network to obtain a sample interference suppression signal: A feature extraction subnetwork extracts features from interfered channel samples to obtain interference channel features. An interference suppression subnetwork performs interference suppression on the interference channel features to obtain interference-suppressed sample channel features. A channel recovery subnetwork performs channel recovery on the interference-suppressed sample channel features to obtain a sample interference suppression channel.

[0035] The interfered signal samples include interfered pilot signals. A pilot signal may be understood as a signal sent in a network for measurement or monitoring purposes.

[0036] The interfered channel samples are obtained by performing channel estimation on the interfered pilot signal.

[0037] Exemplarily, the interference suppression sub-network may include, for example: a recurrent neural network (RNN for short) and a long short-term memory network (LSTM for short).

[0038] The sample interference suppression signal includes a sample interference suppression channel.

[0039] The initial neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a channel recovery subnetwork.

[0040] On the one hand, interference is related to channel estimation. On the other hand, there is the following situation: in the uplink communication direction, for example, transmitted from the user equipment to the network, the signal interference is mainly channel interference. Therefore, the embodiment of the present application can perform channel interference suppression. Specifically, the interference channel characteristics obtained by the feature extraction subnetwork can characterize the characteristics of the interfered channel. The interference suppression subnetwork can perform interference suppression based on the interference channel characteristics to obtain sample channel characteristics after interference suppression. The sample channel characteristics after interference suppression can be restored to a sample interference suppression channel through the channel recovery subnetwork. The sample interference suppression channel is interference suppressed, so channel interference suppression can be achieved.

[0041] Of course, in the downlink communication direction, for example, transmitted from the network to the user equipment, channel interference suppression can also be performed using the communication interference suppression method of the embodiment of the present application.

[0042] In a possible embodiment, for example, the following embodiment can be used to implement a specific example of determining a feedback value based on a sample interference suppression signal and a non-interference signal sample: determining a feedback value based on a sample interference suppression channel and a non-interference channel sample.

[0043] The interference-free channel samples are obtained by performing channel estimation on the interference-free signal samples, and the interference-free signal samples include the interference-free channel samples.

[0044] Exemplarily, the feedback value may be determined according to a difference between samples of the interference-suppressed channel and samples of the interference-free channel.

[0045] In the embodiment of the present application, when performing channel interference suppression, interference-free channel samples are used as labels. The feedback values ​​determined based on the sample interference suppression channels and the interference-free channel samples can accurately characterize the channel interference suppression performance of the current initial neural network. Subsequently, the parameters of the initial neural network can be accurately adjusted based on the feedback values, thereby obtaining a neural network with better channel interference suppression performance.

[0046] In one possible embodiment, for example, the following embodiments can be used to implement a specific example of processing interfered signal samples using an initial neural network to obtain a sample interference suppression signal: A feature extraction subnetwork extracts features from interfered channel samples and interfered data signals to obtain interference channel features and interference data features, respectively. An interference estimation result is determined based on the interference channel features and interference-free channel features. The interference suppression subnetwork performs interference suppression on the interference data features based on the interference estimation result to obtain interference-suppressed sample data features. The data recovery subnetwork performs data recovery on the interference-suppressed sample data features to obtain sample interference suppression data.

[0047] The interfered signal samples include the interfered pilot signal and the interfered data signal. The interfered channel samples are obtained by performing channel estimation on the interfered pilot signal.

[0048] The initial neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a data recovery subnetwork.

[0049] In the downlink communication direction of, for example, a network to a user device, it is typically used to transmit data, so more attention is paid to the accuracy of data transmission. According to the communication interference suppression method of an embodiment of the present application, it can be used for data interference suppression. Specifically, the interference channel characteristics and interference data characteristics obtained by the feature extraction subnetwork can respectively characterize the characteristics of the interfered channel and the interfered data, and the interference estimation results obtained by the interference channel characteristics and the interference-free channel characteristics can actually characterize the channel interference. Since the data signal is transmitted through the channel, the interference to the data signal is actually consistent with the channel interference. According to the interference estimation result (i.e., channel interference), the interference data characteristics can be accurately and efficiently suppressed for communication interference.

[0050] In a possible embodiment, for example, the following embodiment can be used to implement a specific example of determining a feedback value based on sample interference suppression signals and interference-free signal samples: determining a feedback value based on sample interference suppression data and interference-free data samples.

[0051] The interference-free signal samples include interference-free data samples.

[0052] For example, the feedback value may be determined based on a difference between the sample interference suppression data and the interference-free data sample.

[0053] In the embodiment of the present application, when performing data signal interference suppression, interference-free data samples are used as labels. The feedback value determined based on the sample interference suppression data and the interference-free data samples can accurately characterize the data interference suppression performance of the current initial neural network. Subsequently, the parameters of the initial neural network can be accurately adjusted based on the feedback value, thereby obtaining a neural network with better data interference suppression performance.

[0054] In a possible embodiment, for example, the method may further include: sending a training mode end instruction.

[0055] The training mode end instruction is used to instruct the first communication subject to end the neural network training. The training mode end instruction is also used to instruct the second communication subject to send a signal to the first communication subject.

[0056] According to the communication interference suppression method of the embodiment of the present application, the first communication subject can instruct the first communication subject to end the neural network training by sending a training mode end instruction, and the second communication subject and the interference source can work normally. Therefore, the communication interference suppression method of the embodiment of the present application can collect training samples (training samples include interfered signal samples and non-interference signal samples) through the instruction interaction between the first communication subject, the second communication subject and the interference source, perform neural network training according to the training samples, and perform communication interference suppression normally after the neural network training is completed. This series of operations can be performed in sequence, with better real-time performance and higher communication interference suppression accuracy and efficiency.

[0057] In one possible embodiment, for example, the following embodiments can be used to implement wireless communication interference suppression on signals using a neural network to obtain an interference-suppressed signal: A feature extraction subnetwork is used to extract features from a working channel to obtain channel features. An interference suppression subnetwork is used to perform interference suppression on the channel features to obtain interference-suppressed channel features. A channel recovery subnetwork is used to perform channel recovery on the interference-suppressed channel features to obtain an interference-suppressed channel.

[0058] The signal includes a pilot signal. The working channel is obtained by performing channel estimation on the pilot signal.

[0059] The neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a channel recovery subnetwork.

[0060] The interference suppression signal includes an interference suppression channel.

[0061] According to another embodiment of the present disclosure, a communication interference suppression method can be implemented using a neural network to suppress wireless communication interference on a signal, for example, using the following embodiments to obtain a specific example of an interference suppression signal: a feature extraction subnetwork is used to extract features of a working channel and a data signal to obtain channel features and data features, respectively. Interference estimation is performed based on the channel features and the interference-free channel features to obtain a channel interference estimation result. An interference suppression subnetwork is used to suppress interference on the data features based on the channel interference estimation result to obtain interference-suppressed data features. A data recovery subnetwork is used to recover the interference-suppressed data features to obtain interference-suppressed data.

[0062] The signal includes a pilot signal and a data signal, and the working channel is obtained by performing channel estimation on the pilot signal.

[0063] The neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a data recovery subnetwork.

[0064] The interference-free channel feature is obtained by performing channel estimation and feature extraction on the interference-free signal, where the interference-free signal comes from the second communication subject.

[0065] The interference suppression signal includes interference suppression data.

[0066] In a second aspect, the present application relates to a communication interference suppression device, comprising: a signal acquisition module and an interference suppression module.

[0067] The signal acquisition module is configured to acquire a signal from the second communication subject.

[0068] The interference suppression module is configured to use a neural network to perform wireless communication interference suppression on the signal to obtain an interference suppression signal.

[0069] In a third aspect, the present application relates to an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the communication interference suppression method as described above is executed.

[0070] In a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein when the computer instructions are executed, the above-mentioned communication interference suppression method is executed.

[0071] In a fifth aspect, the present application relates to a computer program product, comprising a computer program, wherein the computer program is stored on at least one of a readable storage medium and an electronic device, and when the computer program is executed by a processor, the communication interference suppression method as described above is executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] The following is an introduction to the drawings used in the embodiments of this application.

[0073] 1A and 1B schematically illustrate a schematic diagram of a frequency reuse technology;

[0074] FIG1C schematically shows a schematic diagram of communication interference suppression using interference suppression combining;

[0075] FIG2A schematically shows a schematic diagram of a system architecture of a communication interference suppression method and apparatus according to an embodiment of the present disclosure;

[0076] FIG2B schematically shows another system architecture of a method and apparatus for suppressing communication interference according to an embodiment of the present disclosure;

[0077] FIG2C schematically shows a schematic diagram of another system architecture of the communication interference suppression method and apparatus according to an embodiment of the present disclosure;

[0078] FIG3A schematically shows a flow chart of a communication interference suppression method according to an embodiment of the present disclosure;

[0079] FIG3B schematically shows a schematic diagram of the communication interference suppression method according to an embodiment of the present disclosure in terms of communication interference estimation accuracy;

[0080] FIG3C schematically shows a schematic diagram of the communication interference suppression method in terms of signal-to-noise ratio according to an embodiment of the present disclosure;

[0081] FIG3D schematically shows a schematic diagram of the communication interference suppression method in terms of throughput according to an embodiment of the present disclosure;

[0082] FIG4A schematically shows a schematic diagram of a communication interference suppression method according to another embodiment of the present disclosure;

[0083] FIG4B schematically shows a schematic diagram of a communication interference suppression method according to an embodiment of the present disclosure, when the first communication subject is a target base station and the second communication subject is a target user equipment;

[0084] FIG4C schematically shows a schematic diagram of a communication interference suppression method according to an embodiment of the present disclosure, when the second communication subject is a target base station and the first communication subject is a target user equipment;

[0085] FIG5A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure;

[0086] FIG5B schematically shows a schematic diagram of channel estimation without interference and channel estimation with interference;

[0087] FIG6 schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure;

[0088] FIG7A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure;

[0089] FIG7B schematically shows a schematic diagram of a communication interference suppression method according to an embodiment of the present disclosure applied to data interference suppression;

[0090] FIG8A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure;

[0091] FIG8B schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure;

[0092] FIG9 schematically shows a block diagram of a communication interference suppression device according to an embodiment of the present disclosure; and

[0093] FIG10 schematically shows a block diagram of an electronic device that can implement the communication interference suppression method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0094] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0095] The term "and / or" in this article is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0096] In the description and claims of the embodiments of this application, the terms "first" and "second" are used to distinguish different objects, rather than to describe a specific order of objects. For example, the terms "first target object" and "second target object" are used to distinguish different objects, rather than to describe a specific order of objects.

[0097] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0098] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0099] The following describes in detail the background of the communication interference suppression method according to the embodiment of the present disclosure.

[0100] Wireless Communication

[0101] Wireless communication, also known as wireless communication technology, is a communication method that uses electromagnetic waves such as radio waves or light waves to transmit information.

[0102] Wireless communication generally involves the following processes: analog signal sampling, signal modulation, and signal demodulation. Analog signal sampling converts the analog signal into a digital signal, obtaining the original baseband signal. Signal modulation converts the low-frequency original baseband signal into a high-frequency signal suitable for wireless transmission. The modulated signal can be transmitted over the wireless network. The receiver can receive the modulated signal and perform signal demodulation to recover the original baseband signal.

[0103] Cellular Mobile Communication System

[0104] The cellular mobile communication system is a mobile communication hardware architecture. It is named because the signal coverage of each communication base station that constitutes the system is hexagonal, making the entire coverage network look like a honeycomb.

[0105] In cellular mobile communication systems, the signal coverage area is divided into cells. Each cell can be hexagonal, square, circular, etc. Each cell has a corresponding base station and is also assigned a corresponding frequency band. Adjacent cells have different frequency bands to avoid co-channel interference.

[0106] Cellular mobile communication systems continue to evolve, from the first generation (1G) to the increasingly mature fifth generation (5G). Fifth generation cellular communication systems utilize Multiple-Input Multiple-Output (MIMO) technology to mitigate channel fading during communication, thereby increasing system channel capacity, signal coverage, and transmission rates. In particular, Massive MIMO technology can improve system capacity and spectrum efficiency.

[0107] Multiple Input Multiple Output (MIMO)

[0108] MIMO technology uses multiple transmit and receive antennas at both the transmitter and receiver ends, allowing signals to be transmitted and received via these multiple antennas, thereby improving communication quality. MIMO technology fully utilizes spatial resources, enabling multiple transmissions and multiple receptions through multiple antennas. This can exponentially increase system channel capacity without increasing spectrum resources or antenna transmit power.

[0109] Massive Multiple Input Multiple Output (MMIO)

[0110] Massive Multiple-Input Multiple-Output (MIMO) technology uses dozens or even hundreds of antennas to expand the traditional MIMO antenna system into a massive antenna matrix. This massive antenna matrix uses beamforming technology to focus the energy of transmitted and received signals to a limited area, improving energy efficiency and transmission distance. MIMO's spatial multiplexing technology also improves transmission efficiency.

[0111] Cell-free massive multiple-input multiple-output (mMIMO) systems

[0112] Traditional massive MIMO systems are cell-centric, which is limited by inter-cell interference. Cell-free massive MIMO systems are distributed massive MIMO technologies that are user-centric.

[0113] In cell-free mMIMO, many distributed APs are connected to a central processing unit (CPU), which simultaneously serve all user equipment (UE) in the network.

[0114] An AP can be thought of as a circuit consisting of antenna elements and the signal processing units required to operate them, such as filters, analog-to-digital and digital-to-analog converters (ADCs and DACs), etc. This setup facilitates performing computations locally.

[0115] The following description will be made using a cellular mobile communication system as an example.

[0116] With the substantial growth in communication demand, the number of users served by a base station within the same frequency band has increased significantly, which will cause more serious interference between users within the cell. When adjacent cells share the same frequency band, there is still a certain amount of neighboring cell interference.

[0117] In particular, under the architecture of the massive multiple-input multiple-output technology of the future sixth-generation cellular communication system (i.e. 6G), interference within the cell will increase. In addition, the deployment of future non-cellular massive multiple-input multiple-output systems will also bring more interference to neighboring cells, making how to accurately and efficiently suppress wireless communication interference a technical problem that needs to be solved urgently.

[0118] As shown in FIG1A and FIG1B , in some implementations, frequency reuse technology is used to physically separate base stations using the same frequency, and different spectrum resources are allocated to neighboring base stations to reduce co-channel interference in neighboring cells to an allowable range.

[0119] For example, in the example of Figure 1A, among the multiple base stations in the cellular structure including the base station BS_1 and the base station BS_2, the cells of the adjacent base stations BS_1 and BS_2 include multiple user devices (Figure 1B schematically shows three user devices in the cell of the base station BS_1: user equipment UE_11 to user equipment UE_13, and Figure 1B also schematically shows three user devices in the cell of the base station BS_2: user equipment UE_21 to user equipment UE_23). When the user devices in the adjacent cells communicate, the communication interference can be reduced because different spectrum resources are used between the adjacent cells.

[0120] However, due to the limited frequency band resources, this implementation method uses frequency reuse, which will result in a reduction in the spectrum resources allocated to a single base station, or the base station needs to request the required spectrum resources, resulting in low spectrum resource utilization and reduced spectrum efficiency of the network.

[0121] As shown in Figure 1C, in some implementations, interference rejection combining (IRC) is employed to whiten colored interference to obtain the optimal signal reception direction W, where the signal is subject to minimal interference. Interference whitening can be understood as the process of converting interference with a specific distribution into a uniform distribution. For example, the optimal signal reception direction W can be obtained using the following formula (1-1).

[0122] Among them, H0 T Represents the transpose of the channel estimate, H0 represents the channel estimate, P i Characterizes the interference power, P x Characterizes the signal power, represents the noise power, and I represents the unit matrix of the noise effect.

[0123] In MIMO scenarios, interference suppression combining suffers from interference estimation errors, resulting in poor performance in real-world scenarios with multiple interference sources and real-time variations. Furthermore, in scenarios where the receiver has only a single antenna, the minimum mean square error (MMSE) equalization matrix obtained through IRC is one-dimensional, making it impossible to determine the optimal signal reception direction W. This approach degenerates into a traditional maximum ratio combining (MRC) receiver, resulting in no performance gain.

[0124] In some implementations, interference suppression is performed using successive interference cancellation (SIC). For example, the signal y(t) received at time t can be obtained using the following formula (1-2). Energy signals in y(t) can be separated one by one from high to low energy, and signals belonging to different users can be gradually separated. Channel estimation can then be performed based on each user's signal, and interference suppression can be performed based on the channel estimation results. Effective interference suppression can be achieved through multiple iterations.

[0125] Among them, x i (t) represents the signal sent by the i-th user at time t, and n(t) represents the noise at time t.

[0126] This approach assumes that interference comes from a known number of users and requires that the type of signal transmitted by any user be known. In reality, in scenarios with neighboring cell interference, the type of interference signal from neighboring users cannot be directly determined due to unknown scheduling conditions. Therefore, additional signaling is required to determine the characteristics and quantity of the interfering signal, resulting in high signaling overhead.

[0127] The embodiments of the present disclosure provide an interference suppression technology to improve the traditional interference suppression processing method.

[0128] Figures 2A to 2C schematically illustrate a system architecture of a method and apparatus for suppressing communication interference according to an embodiment of the present disclosure. It should be noted that Figures 2A to 2C are merely examples of system architectures to which the method and apparatus for suppressing communication interference according to an embodiment of the present disclosure may be applied, to help those skilled in the art understand the technical content of the present disclosure, but do not imply that the embodiments of the present disclosure may not be applied to other devices, systems, environments, or scenarios.

[0129] FIG2A schematically illustrates a system architecture of a method and apparatus for suppressing communication interference according to an embodiment of the present disclosure.

[0130] As shown in FIG. 2A , the system architecture in the embodiment of the present disclosure may include: a first communication subject 201 , a second communication subject 202 , and an interference source 203 .

[0131] For the wireless communication between the second communication subject 202 and the first communication subject 201, the second communication subject 202 can send a signal to the first communication subject 201, and the first communication subject 201 can receive the signal. Ideally, the signal received by the first communication subject 201 is the same as the signal sent by the second communication subject 202. In real-world scenarios, due to the presence of an interference source 203, the interference source 203 can interfere with the signal X sent by the second communication subject 202, causing the first communication subject 201 to receive a distorted signal X, affecting communication efficiency. Therefore, communication interference suppression is necessary.

[0132] The first communication subject 201 may perform wireless communication interference suppression on the received signal Y to obtain an interference suppression signal.

[0133] The first communication subject and the second communication subject can be understood as device subjects supporting wireless communication. The interference source can be understood as a device subject that interferes with the signal when the second communication subject sends a signal to the first communication subject.

[0134] Exemplarily, the first communication subject, the second communication subject and the interference source may respectively include: a communication component, a baseband processing unit (Base Band Unit, referred to as BBU), a chip module, a base station, and a user equipment within the cell of the base station.

[0135] FIG2B and FIG2C schematically illustrate the system architecture of a communication interference suppression method and apparatus according to another embodiment of the present disclosure, respectively.

[0136] As shown in Figure 2B, the first communication subject 201 is the target base station BS_t, the second communication subject 202 is the target user equipment UE_t in the cell of the target base station BS_t, and the interference source 203, for example, includes at least one of the following: the neighboring base station BS_a of the target base station BS_t and other user equipment UE_e in the cell of the target base station BS_t.

[0137] As shown in Figure 2C, the second communication subject 202 is the target base station BS_t, the first communication subject 201 is the target user equipment UE_t in the cell of the target base station BS_t, and the interference source 203 may, for example, include at least one of the following: the neighboring base station BS_a of the target base station BS_t and other user equipment UE_e in the cell of the target base station BS_t.

[0138] It should be noted that the communication interference suppression method provided in the embodiments of the present disclosure can generally be performed by the first communication subject 201. Accordingly, the communication interference suppression apparatus provided in the embodiments of the present disclosure can generally be provided in the first communication subject 201. The communication interference suppression method provided in the embodiments of the present disclosure can also be performed by another communication subject that is different from the first communication subject 201 and that is capable of wirelessly communicating with the second communication subject 202. Accordingly, the communication interference suppression apparatus provided in the embodiments of the present disclosure can also be provided in another communication subject that is different from the first communication subject 201 and that is capable of wirelessly communicating with the first communication subject 201.

[0139] The number of the first communication subjects 201, the second communication subjects 202, and the interference sources 203 in the above embodiment is only for illustration, and any number of the first communication subjects 201, the second communication subjects 202, and the interference sources 203 can be set as required.

[0140] The present disclosure provides a method for suppressing communication interference. The following describes the communication interference suppression method according to an exemplary embodiment of the present disclosure, in conjunction with the system architecture of Figures 2A to 2C and with reference to Figures 3A to 8B. The communication interference suppression method according to the present disclosure embodiment can be executed, for example, by the first communication subject 201 shown in Figures 2A to 2C.

[0141] FIG3A schematically shows a flow chart of a communication interference suppression method 300 according to an embodiment of the present disclosure.

[0142] As shown in FIG3A , a communication interference suppression method 300 according to an embodiment of the present disclosure is applied to a first communication subject. The communication interference suppression method 300 includes operations S310 to S320 .

[0143] In operation S310, a signal from a second communication subject is acquired.

[0144] The signal obtained by the first communication subject from the second communication subject is interfered with. The wireless communication interference may come from some interference sources, or may be loss during signal transmission (loss may include noise, for example).

[0145] Of course, in some cases, the signal received by the first communication subject from the second communication subject is not interfered with, and the signal can still be processed by the communication interference suppression method of the embodiment of the present disclosure.

[0146] In operation S320, wireless communication interference suppression is performed on the signal using a neural network to obtain an interference suppression signal.

[0147] It should be noted that the neural network in the embodiment of the present disclosure is trained and can be directly used for wireless communication interference.

[0148] Neural Network (NN) can be understood as a machine learning model. A neural network can be composed of neural units. A neural unit can refer to a s The output of the operation unit with the sum of intercept 1 as input can be expressed by the following formula (1-3).

[0149] Where, s = 1, 2, ... n, n is a natural number greater than 1, W s is x s The weight of the neural unit is , and b is the bias of the neural unit. f is the activation function of the neural unit, which is used to introduce nonlinear characteristics into the neural network to convert the input signal of the neural unit into the output signal. The output signal of the activation function can be used as the input of the next convolutional layer. The activation function can be, for example, a sigmoid function.

[0150] A neural network is formed by connecting many of these single neural units. That is, the output of one neural unit can be the input of another. The input of each neural unit can be connected to the local receptive field of the previous layer to extract features of the local receptive field, which can be an area composed of several neural units.

[0151] The communication interference suppression method of the embodiment of the present disclosure uses a neural network to suppress wireless communication interference on the signal. It only takes the received signal as input and does not pay attention to what interference source the signal is interfered with. Therefore, there is no need to solve each interference source through calculation, nor is there any need to calculate the interference distribution of the signal. The signal only needs to be input into the neural network to suppress the interference of the signal. This can greatly reduce the complexity of communication interference suppression and has higher interference suppression efficiency. It is particularly suitable for complex scenarios such as multiple input and multiple output and multiple interference sources.

[0152] On the one hand, in a multi-input and multi-output scenario, especially in a scenario where there are many interference sources and the interference changes in real time, the communication interference suppression method of the embodiment of the present disclosure can still use the signal at each moment as input regardless of how the interference changes in real time, without paying attention to the interference source, that is, there is no need to solve each interference source. In a scenario where there are a large number of interference sources, the error in solving the interference source can be greatly reduced, and there is no need to perform interference distribution calculation in real time, which can greatly reduce the error in real-time calculation of interference distribution, thereby improving the accuracy and efficiency of interference suppression. In addition, this process will not interfere with the normal operation of the communication system, and therefore will not affect the performance of the communication system. For example, the communication interference suppression method of the embodiment of the present disclosure will not interfere with the normal operation of the communication system based on multi-input and multi-output technology, and therefore will not reduce the performance gain of the communication system.

[0153] On the other hand, the communication interference suppression method of the embodiment of the present disclosure does not require real-time and frequent acquisition of neighboring cell scheduling information through signaling, and does not require the acquisition of the signal type of the user equipment in the neighboring cell. Therefore, there is no signaling overhead for acquiring the signal type of the user equipment in the neighboring cell. As a result, the communication interference suppression method of the embodiment of the present disclosure has higher interference suppression efficiency.

[0154] On the other hand, the trained neural network can ensure a certain degree of interference suppression accuracy. Therefore, the interference suppression signal obtained by using the neural network to suppress wireless communication interference is closer to the original signal sent by the transmitter. The communication interference suppression method of the disclosed embodiment has higher accuracy for wireless communication interference suppression and can improve wireless communication efficiency.

[0155] On the other hand, taking wireless communication between cells of base stations in a cellular system as an example, the communication interference suppression method of the embodiment of the present disclosure only focuses on the received signal. In the process of wireless communication interference suppression, it does not affect the spectrum resources of each cell, and therefore does not reduce the spectrum efficiency of the network.

[0156] In summary, according to the communication interference suppression method of the embodiment of the present disclosure, by inputting the signal into the neural network, an interference suppression signal is obtained, which can adapt to large-scale multiple-input multiple-output technology, cell-free multiple-input multiple-output technology, etc. Since 5G and 6G technologies mostly adopt large-scale multiple-input multiple-output technology and cell-free multiple-input multiple-output technology, the communication interference suppression method of the embodiment of the present disclosure can be adapted to network architecture deployments such as 5G, 6G, etc.

[0157] In addition, the communication interference suppression method according to the embodiment of the present disclosure has better performance in various aspects such as communication interference estimation accuracy, signal-to-noise ratio, and throughput.

[0158] Figure 3B schematically illustrates the communication interference estimation accuracy of the communication interference suppression method according to an embodiment of the present disclosure. In the example of Figure 3B , the horizontal axis represents the number of antennas, and the vertical axis represents the communication interference estimation error (this error can be, for example, the normalized mean squared error, or NMSE). White columns represent the communication interference estimation error using the communication interference suppression method according to an embodiment of the present disclosure, and gray columns represent the communication interference estimation error when the communication interference suppression method according to an embodiment of the present disclosure is not used.

[0159] Figure 3C schematically illustrates a schematic diagram of the communication interference suppression method according to an embodiment of the present disclosure in terms of signal-to-noise ratio. In the example of Figure 3C , the horizontal axis represents the signal-to-noise ratio (SINR), the vertical axis represents the block error rate (BLER), L1 represents the signal-to-noise ratio when the communication interference suppression method according to an embodiment of the present disclosure is not used, and L2 represents the signal-to-noise ratio when the communication interference suppression method according to an embodiment of the present disclosure is used.

[0160] Figure 3D schematically illustrates a schematic diagram of the throughput of the communication interference suppression method according to an embodiment of the present disclosure. In the example of Figure 3D, the vertical axis represents the throughput gain (in megabits per second, or Mbps), and the four gray columns C1 to C4 represent the throughput without interference, the throughput using the communication interference suppression method according to an embodiment of the present disclosure, the throughput using Maximum Ratio Combining (MRC), and the throughput using Interference Rejection Combining (IRC), respectively.

[0161] Exemplarily, the neural network may include: a deep neural network, a convolutional neural network, a recurrent neural network, etc.

[0162] (1) Deep neural network (DNN)

[0163] Deep neural networks, also known as multi-layer neural networks, can be understood as neural networks with many hidden layers. The "many" here does not have a specific metric. Based on the position of different layers in DNN, the neural network inside DNN can be divided into three categories: input layer, hidden layer, and output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and the layers in between are all hidden layers. The layers are fully connected, that is, any neuron in the i-th layer must be connected to any neuron in the i+1-th layer. Although DNN looks complicated, the work of each layer is actually not complicated. Simply put, it can be expressed as the following linear relationship: in, is the input vector, is the output vector, is the offset vector, W is the weight matrix (also called coefficient), and α() is the activation function. Each layer is just an input vector After such a simple operation, the output vector Since there are many DNN layers, the coefficient W and the offset vector The definition of these parameters in DNN is as follows: Take the coefficient W as an example: Assume that in a three-layer DNN, the linear coefficient from the 4th neuron in the second layer to the 2nd neuron in the third layer is defined as The superscript 3 represents the layer number of the coefficient W, while the subscript corresponds to the output of the third layer index 2 and the input of the second layer index 4. In summary, the coefficient from the kth neuron in the L-1th layer to the jth neuron in the Lth layer is defined as It's important to note that the input layer has no W parameter. In deep neural networks, more hidden layers allow the network to better capture complex real-world situations. Theoretically, a model with more parameters has higher complexity and greater "capacity," meaning it can handle more complex learning tasks. The process of training a deep neural network is essentially the process of learning the weight matrix, with the ultimate goal of obtaining the weight matrices for all layers of a trained deep neural network (a weight matrix formed by the vectors W across many layers).

[0164] (2) Convolutional neural network (CNN)

[0165] A convolutional neural network (CNN) is a deep neural network with a convolutional structure and a deep learning architecture. A deep learning architecture involves multiple levels of learning at different levels of abstraction using machine learning algorithms. As a deep learning architecture, a CNN is a feed-forward artificial neural network whose neurons respond to input images. A CNN consists of a feature extractor consisting of convolutional and pooling layers. The feature extractor can be thought of as a filter, and the convolution process can be thought of as convolving an input image or feature map with a trainable filter.

[0166] A convolutional layer is a layer of neurons in a convolutional neural network that performs convolution on the input signal. A convolutional layer can include multiple convolution operators, also known as kernels. In image processing, a convolution operator acts as a filter that extracts specific information from the input image matrix. A convolution operator is essentially a weight matrix, which is usually predefined. During the convolution operation, the weight matrix is ​​typically applied horizontally to the input image, pixel by pixel (or two pixels by two pixels, depending on the stride), to extract specific features from the image. The size of the weight matrix should be proportional to the image size. It is important to note that the depth dimension of the weight matrix is ​​the same as the depth dimension of the input image; during the convolution operation, the weight matrix extends across the entire depth of the input image. Therefore, convolution with a single weight matrix produces a convolved output with a single depth dimension. However, in most cases, multiple weight matrices of the same size (row × column) are applied instead. This is known as multiple homogeneous matrices. The outputs of each weight matrix are stacked to form the depth dimension of the convolutional image. The dimension here can be understood as being determined by the "multiple" mentioned above. Different weight matrices can be used to extract different features from the image. For example, one weight matrix is ​​used to extract image edge information, another weight matrix is ​​used to extract specific colors in the image, and yet another weight matrix is ​​used to blur unwanted noise in the image. The multiple weight matrices have the same size (rows × columns), and the feature maps extracted by these multiple weight matrices of the same size are also the same size. The extracted feature maps of the same size are then merged to form the output of the convolution operation. In practical applications, the weight values ​​in these weight matrices require extensive training. The weight matrices formed by the trained weight values ​​can be used to extract information from the input image, allowing the convolutional neural network to make accurate predictions. When a convolutional neural network has multiple convolutional layers, the initial convolutional layers often extract more general features, which can also be called low-level features. As the depth of the convolutional neural network increases, the features extracted by subsequent convolutional layers become increasingly complex, such as high-level semantic features. Features with higher semantics are more applicable to the problem being solved.

[0167] Because it's often necessary to reduce the number of trainable parameters, pooling layers are often periodically introduced after convolutional layers. This can be done in a single convolutional layer followed by a pooling layer, or in a multi-layered system followed by one or more pooling layers. In image processing, the sole purpose of a pooling layer is to reduce the spatial size of an image. Pooling layers can include average pooling and / or max pooling operators, which are used to downsample the input image to produce a smaller image. The average pooling operator calculates the average value of pixel values ​​within a specific range, producing the average pooling result. The max pooling operator takes the pixel with the largest value within a specific range as the max pooling result. Furthermore, just as the size of the weight matrix used in a convolutional layer should be related to the image size, the operators in a pooling layer should also be related to the image size. The output image size after processing by a pooling layer can be smaller than the size of the image input to the pooling layer. Each pixel in the output image represents the average or maximum value of the corresponding subregion of the image input to the pooling layer.

[0168] After being processed by the convolutional layer / pooling layer, the convolutional neural network is not sufficient to output the required output information. Because as mentioned above, the convolutional layer / pooling layer only extracts features and reduces the parameters brought by the input image. However, in order to generate the final output information (the required class information or other related information), the convolutional neural network needs to use the neural network layer to generate one or a group of outputs of the required number of classes. Therefore, the neural network layer may include multiple hidden layers, and the parameters contained in the multiple hidden layers can be pre-trained based on relevant training data of a specific task type. For example, the task type may include communication interference suppression.

[0169] Optionally, after the multiple hidden layers in the neural network layer, an output layer of the entire convolutional neural network is also included. The output layer has a loss function similar to the classification cross entropy, which is specifically used to calculate the prediction error. Once the forward propagation of the entire convolutional neural network is completed, the backpropagation will begin to update the weight values ​​and biases of the aforementioned layers to reduce the loss of the convolutional neural network and the error between the result output by the convolutional neural network through the output layer and the ideal result.

[0170] (3) Recurrent neural networks (RNN)

[0171] Recurrent neural networks are designed to process sequential data. In traditional neural network models, layers are fully connected, from the input layer to the hidden layer to the output layer, while nodes within each layer are disconnected. While these conventional neural networks have solved many difficult problems, they are still inadequate for many others. For example, predicting the next word in a sentence generally requires using previous words, as the previous and subsequent words in a sentence are not independent. RNNs are called recurrent neural networks because the current output of a sequence is dependent on the previous output. Specifically, the network remembers previous information and applies it to the calculation of the current output. This means that nodes within the hidden layer are no longer disconnected but connected, and the input to a hidden layer includes not only the output of the input layer but also the output of the previous hidden layer. In theory, RNNs can process sequences of any length. Training RNNs is similar to training traditional CNNs or DNNs, using the same backpropagation algorithm. However, there is a key difference: when an RNN is unrolled, its parameters, such as W, are shared, unlike the traditional neural network mentioned above. Furthermore, when using the gradient descent algorithm, the output of each step depends not only on the network at the current step, but also on the state of the network at the previous steps. This learning algorithm is called Back Propagation Through Time (BPTT).

[0172] Since we already have convolutional neural networks, why do we still need recurrent neural networks? The reason is simple. Convolutional neural networks assume that elements are independent of each other, and that inputs and outputs are also independent, such as cats and dogs. However, in the real world, many elements are interconnected, such as the changes in stock prices over time. Or, for example, someone says, "I love traveling, and my favorite place is Yunnan. I must visit it someday." Humans should know to fill in the blank with "Yunnan." This is because humans make inferences based on context, but how can machines do the same? That's where RNNs come in. RNNs aim to give machines the ability to remember, like humans do. Therefore, the output of an RNN depends on both the current input and historical memory.

[0173] FIG4A schematically shows a schematic diagram of a communication interference suppression method according to another embodiment of the present disclosure.

[0174] As shown in FIG4A , the communication interference suppression method according to another embodiment of the present disclosure further includes operation S401 , for example, before acquiring a signal from a second communication subject.

[0175] In operation S401 , a training mode start instruction is sent to start neural network training.

[0176] The training mode start instruction is used to instruct the interference source, the first communication subject, and the second communication subject to start wireless communication in a non-interference mode.

[0177] In the non-interference mode, the second communication subject obtains the first signal that is not interfered by the interference source.

[0178] For example, the first communication subject may send a training mode start instruction to the second communication subject and the interference source to start neural network training. After receiving the training mode start instruction, the second communication subject may send a non-interference signal that is not interfered with by the interference source to the first communication subject.

[0179] Exemplarily, when the interference source receives a training mode start-up instruction, the interference source can, for example, send signal S1, and the second communication subject can, for example, send signal S2. Signal S1 can, for example, be orthogonal to signal S2 or in a different frequency band. This can ensure that the signal S1 of the interference source does not interfere with the signal S2 sent by the second communication subject, and the second communication subject can obtain the first signal that is not interfered with by the interference source.

[0180] Exemplarily, when the first communication subject is a target base station, the second communication subject is a target user device within the cell of the target base station, and the interference source is a neighboring base station and other user devices within the cell of the target base station, for example, the target base station can send a training mode start instruction to the target user device, the neighboring base station, and other user devices to start neural network training.

[0181] For example, where the second communication subject is a target base station, the first communication subject is a target user equipment within the cell of the target base station, and the interference sources are neighboring base stations and other user equipment within the cell of the target base station, the target user equipment may send a training mode start instruction to the target base station, the neighboring base stations, and the other user equipment to start neural network training. Alternatively, the target base station may send a training mode start instruction to the target user equipment, the neighboring base stations, and the other user equipment, and the target user equipment may start neural network training after receiving the training mode start instruction.

[0182] For example, the training mode activation instruction may include instruction data and information data. The instruction data may be used to instruct the first communication subject, the second communication subject, and the interference source to activate the non-interference mode. The information data may be used to indicate the operating time, operating duration, and other information of the non-interference mode. The information data may be configurable.

[0183] For example, the working time, working duration and other information of the non-interference mode may be preset, that is, the working time of the non-interference mode may be timed.

[0184] The communication interference suppression method of the embodiment of the present disclosure can realize the initiation of neural network training and the application of the neural network (the application of the neural network is to use the neural network to suppress wireless communication interference) on the first communication subject by sending a training mode startup instruction to start neural network training. The startup of the neural network model is instructed by the training mode startup instruction. The training mode startup instruction acts on the second communication subject and the interference source, which can realize the interaction of each subject involved in communication interference suppression (each subject includes the first communication subject, the second communication subject, and the interference source), which is closer to the actual communication interference scenario. Therefore, the communication interference suppression method of the embodiment of the present disclosure is more accurate and efficient in suppressing communication interference.

[0185] As shown in FIG4A , the communication interference suppression method according to yet another embodiment of the present disclosure further includes operation S402 .

[0186] In operation S402, an initial neural network is trained using non-interference signal samples and interfered signal samples to obtain a neural network.

[0187] The interference-free signal sample is obtained based on the first signal, and the interfered signal sample is obtained based on the second signal, where the second signal is a signal interfered with by an interference source obtained by the second communication subject.

[0188] As shown in FIG4A , for example, the second communication subject can obtain a first signal that is not interfered with by an interference source in the non-interference mode. The second communication subject can also obtain a second signal that is interfered with by an interference source when the non-interference mode ends. The second communication subject can, for example, send the first signal and the second signal to the first communication subject, and the first communication subject can, for example, determine a non-interference signal sample and an interfered signal sample based on the first signal and the second signal, respectively.

[0189] For example, the first signal and the second signal can be used as interference-free signal samples and interference-affected signal samples, respectively. For example, the first signal and the second signal can be preprocessed to obtain interference-free signal samples and interference-affected signal samples, respectively. The preprocessing here is used to make the first signal and the second signal easier to process by the neural network.

[0190] According to the communication interference suppression method of the embodiment of the present disclosure, the difference between the interference-free signal samples and the interfered signal samples can characterize the interference signal, and the interference signal can be used to suppress communication interference. Therefore, according to the communication interference suppression method of the embodiment of the present disclosure, by using the interference-free signal samples and the interfered signal samples to train the initial neural network, the obtained neural network can be used to accurately estimate the interference signal and suppress communication interference.

[0191] As shown in FIG4A , the communication interference suppression method according to yet another embodiment of the present disclosure may further include, for example: sending a training mode end instruction.

[0192] The training mode end instruction is used to instruct the first communication subject to end the neural network training. The training mode end instruction is also used to instruct the second communication subject to send a signal to the first communication subject.

[0193] For example, the first communication subject may send a training mode end instruction to the second communication subject and the interference source. After the second communication subject and the interference source receive the training mode end instruction, the second communication subject sends a signal to the first communication subject, and the interference source will normally interfere with the signal.

[0194] According to the communication interference suppression method of the embodiment of the present disclosure, the first communication subject can instruct the first communication subject to end the neural network training by sending a training mode end instruction, and the second communication subject and the interference source can work normally. Therefore, the communication interference suppression method of the embodiment of the present disclosure can collect training samples (training samples include interfered signal samples and non-interference signal samples) through the instruction interaction between the first communication subject, the second communication subject and the interference source, perform neural network training based on the training samples, and can also perform communication interference suppression normally after the neural network training is completed. This series of operations can be executed in sequence, with better real-time performance and higher communication interference suppression accuracy and efficiency.

[0195] FIG4B schematically shows a schematic diagram of a communication interference suppression method according to an embodiment of the present disclosure, when the first communication subject is a target base station and the second communication subject is a target user equipment.

[0196] As shown in Figure 2B, the first communication subject is the target base station, and the second communication subject is the target user equipment in the cell of the target base station (in the example of Figure 4B, the user equipment UE includes the target user equipment and other user equipment, UE is User Equipment), and the interference source includes, for example: the neighboring base station of the target base station and other user equipment in the cell of the target base station.

[0197] In the example of Figure 4B, the target base station C sends a training mode start instruction to the neighboring base station and the user equipment UE to start the training mode, and the target base station C performs neural network training in the training mode. The neighboring base station confirms the receipt of the training mode start instruction by feeding back a confirmation character (ACK for short) to the target base station C, and the neighboring base station then enters the non-interference mode. Similar to the neighboring base station, other user equipment enters the non-interference mode. In the non-interference mode, the target base station C cooperates with the neighboring base station and the user equipment UE to collect non-interference signal samples (the non-interference signal samples are obtained based on the first signal, the first signal comes from the target user equipment, and the non-interference signal samples include the non-interference channel estimate H0, and multiple non-interference channel estimates H0 form the {H0} training set). After completing the collection of the non-interference signal samples, the neighboring base station and the user equipment UE exit the non-interference mode (for example, a non-interference mode exit frame can be fed back to the target base station C to indicate the exit from the non-interference mode). The target user equipment can normally send the second signal with interference to the target base station, and the target base station C collects the interfered signal samples (the interfered signal samples are determined according to the second signal, and the interfered signal samples include the interfered channel estimation H I , multiple interfered channel estimates H I Form {H I} training set). The target base station can perform neural network training based on the interference-free signal samples and the interfered signal samples (Figure 4B schematically shows an example of training the initial neural network when the initial neural network is a pre-trained model. The training process is to update the network parameters of the pre-trained model). After the neural network training is completed, the target base station C broadcasts the training mode end instruction and allows the neighboring base stations and user equipment UE to enter the working mode. In the working mode, the target user equipment sends a signal interfered by the interference source to the target base station. The target base station C performs channel estimation based on the received signal and obtains the interfered channel estimate H I , you can also use H I The trained neural network is input and used to suppress wireless communication interference.

[0198] FIG4C schematically shows a schematic diagram of a communication interference suppression method according to an embodiment of the present disclosure, when the second communication subject is a target base station and the first communication subject is a target user equipment.

[0199] As shown in Figure 2C, the second communication subject is the target base station, the first communication subject is the target user equipment in the cell of the target base station (in the example of Figure 4C, the user equipment UE includes the target user equipment and other user equipment, UE is User Equipment), and the interference source includes, for example: the neighboring base station of the target base station and other user equipment in the cell of the target base station.

[0200] In the example of Figure 4C, the target base station C sends a training mode start instruction to the neighboring base station and the user equipment UE to start the training mode, and the target user equipment performs neural network training in the training mode. The neighboring base station confirms the receipt of the training mode start instruction by feeding back a confirmation character (ACK) to the target base station C, and the neighboring base station then enters the non-interference mode. Similar to the neighboring base station, other user equipment enters the non-interference mode. In the non-interference mode, the target user equipment cooperates with the neighboring base station, the target base station C and other user equipment to collect non-interference signal samples (the non-interference signal samples are obtained based on the first signal, the first signal comes from the target base station C, and the non-interference signal samples include the non-interference channel estimate H0, and multiple non-interference channel estimates H0 form the {H0} training set). After completing the collection of the non-interference signal samples, the neighboring base station and the user equipment UE exit the non-interference mode (for example, a non-interference mode exit frame can be fed back to the target base station C to indicate the exit from the non-interference mode). The target base station C can normally send the second signal with interference to the target user equipment, and the target user equipment collects the interfered signal samples (the interfered signal samples are determined according to the second signal, and the interfered signal samples include the interfered channel estimation H I , multiple interfered channel estimates H I Form {H I} training set). The target user equipment can perform neural network training based on the interference-free signal samples and the interfered signal samples (Figure 4C schematically shows an example of training the initial neural network when the initial neural network is a pre-trained model. The training process is to update the network parameters of the pre-trained model). After the neural network training is completed, the target base station C broadcasts the training mode end instruction and allows the neighboring base stations and user equipment UE to enter the working mode. In the working mode, the target base station sends the signal interfered by the interference source to the target user equipment. The target user equipment performs channel estimation based on the received signal and obtains the interfered channel estimate H I In the example of FIG4C , an example of taking H0 from the {H0} training set as prior information is also schematically shown. For example, H I Input the trained neural network, the prior information H0 can be used as the input to f(H I ) is used as a benchmark for comparison to perform communication interference estimation and interference suppression, and f(*) characterizes the processing of the neural network.

[0201] FIG5A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure.

[0202] As shown in FIG5A , according to a communication interference suppression method according to yet another embodiment of the present disclosure, a neural network may be trained by, for example, the following operations S510 to S530 .

[0203] In operation S510 , the interfered signal sample 501 is processed using the initial neural network 502 to obtain a sample interference suppressed signal 503 .

[0204] The initial neural network can be understood as a neural network with initial network parameters. The process of training the initial neural network can adjust the network parameters. The network parameters of the neural network obtained after training make the performance of the neural network for wireless communication interference suppression better.

[0205] Network parameters may include the weights and biases of each network layer / each neuron of the neural network.

[0206] The sample interference suppression signal obtained by processing the disturbed signal sample using the initial neural network can be understood as the result obtained after the neural network with the initial network parameters performs interference suppression.

[0207] In operation S520 , a feedback value 505 is determined based on the sample interference suppression signal 503 and the interference-free signal sample 504 .

[0208] Since the sample interference suppression signal can be understood as the result obtained after interference suppression of a neural network with initial network parameters, the result obtained after interference suppression under ideal conditions should be an undisturbed signal. Therefore, the communication interference suppression method of the embodiment of the present disclosure uses undisturbed interference-free signal samples as labels. Based on the sample interference suppression signal and the interference-free signal sample, the feedback value determined can characterize the interference suppression performance of the current neural network.

[0209] In operation S530 , the parameters of the initial neural network 502 are adjusted according to the feedback value 505 to obtain a neural network 506 .

[0210] Since the feedback value can characterize the interference suppression performance of the current neural network, by adjusting the parameters of the initial neural network according to the feedback value, a neural network with better interference suppression performance can be obtained, thus completing the neural network training.

[0211] Exemplarily, the initial neural network may be a pre-trained neural network.

[0212] A pre-trained neural network can be understood as a neural network obtained through prior training on a large-scale data set. The goal of pre-training is to train a general neural network that can be fine-tuned or transferred using a smaller business data set in subsequent tasks.

[0213] The communication interference suppression method of the disclosed embodiment trains a neural network and utilizes the trained neural network to suppress communication interference. The communication interference suppression method is executed on a first communication subject, such as a user device or a base station, and supports small-scale neural network training. Therefore, higher requirements are placed on the training speed and efficiency of the neural network. The pre-trained neural network has predetermined parameters and only requires fine-tuning of the parameters. Therefore, the training speed and efficiency of the pre-trained neural network are faster and more efficient, which is suitable for the specific scenarios of the communication interference suppression method of the disclosed embodiment.

[0214] The relevant principles of the communication interference suppression method of some embodiments of the present disclosure will be described in detail below.

[0215] The first communication subject receives a signal containing noise and interference, and this signal can be represented by the following formula (1-4).

[0216] Y represents the signal received by the first communication subject, H0 represents the interference-free channel estimation, X represents the signal sent by the second communication subject, I i represents the interference signal generated by the i-th interference source, n0 represents the noise, H I Characterize the channel estimation with interference. Collectively referred to as interference signals.

[0217] Channel estimation can be understood as the process of estimating the characteristics of a channel using the various states exhibited by the received signal. Channel estimation is a mathematical representation of the effect of the channel on the input signal.

[0218] The wireless environment is complex and ever-changing. Signals are subject to various interferences during propagation. When they reach the receiver, the amplitude, phase, and frequency of the signals change significantly. The role of channel estimation and channel equalization is to restore the signal as much as possible.

[0219] The interference-free channel estimation can be understood as the result obtained by performing channel estimation on the interference-free signal, and the interference-impaired channel estimation can be understood as the result obtained by performing channel estimation on the interfered signal.

[0220] The following formula (1-5) can be obtained by transforming the above formula (1-4).

[0221] From the formula (1-5), we can get: the interference signal and the channel estimate without interference H0, the channel estimate with interference H I If the channel estimate H0 without interference and the channel estimate H with interference can be obtained at the same time I , the interference signal can be accurately estimated.

[0222] In fact, with current technology, only the interference-free signal estimate H0 and the interference-containing signal estimate H can be obtained within one frame. I Therefore, it is impossible to simultaneously obtain the interference-free channel estimate H0 and the interference channel estimate H I , determine the interference signal.

[0223] In fact, it is also found that: as shown in Figure 5B, the interference-free channel estimate H0 has characteristics such as flatness in the frequency domain and continuous changes in amplitude and phase in the time domain (the dotted line L3 shown in Figure 5B shows the interference-free channel estimate, and the solid line L4 shown in Figure 5B shows the interference-containing channel estimate. The difference between the two can characterize the interference signal). These characteristics allow the interference-free channel estimate H0 to be separated from the time dimension, that is, there is no need to pay attention to the specific changes in the interference-free channel estimate, and only these characteristics can be used to characterize the interference-free channel estimate H0. In this way, it is not necessary to simultaneously know the interference-free channel estimate H0 and the interference-containing channel estimate H0. I For example, the characteristics of the historical interference-free channel estimate H0 can be used as prior information, and the characteristics of the interference-free channel estimate H0 can be compared with the interference channel estimate H I , the interference signal can be accurately determined, and communication interference suppression can be accurately performed.

[0224] Specifically, for example, the interference channel estimation H I It is also converted into features, that is, the operation of determining the interference signal is transferred into the feature domain. The following formula (1-6) represents the operation of determining the interference signal.

[0225] Here, f(*) represents the processing of the neural network.

[0226] FIG6 schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure.

[0227] As shown in FIG6 , for example, the following embodiments can be used to implement a specific example of using an initial neural network to process interfered signal samples to obtain a sample interference suppression signal.

[0228] In operation S611 , feature extraction is performed on the interfered channel sample 602 according to the feature extraction sub-network N1 to obtain an interference channel feature 603 .

[0229] The interfered signal samples include an interfered pilot signal.

[0230] A pilot signal may be understood as a signal sent in a network for measurement or monitoring purposes.

[0231] Exemplarily, the pilot signal may include a channel sounding reference signal (Sounding Reference Signal, abbreviated as SRS) and a demodulation reference signal (Demodulation reference signal, abbreviated as DMRS).

[0232] The interfered channel samples are obtained by performing channel estimation on the interfered pilot signal.

[0233] In the example of FIG6 , a specific example of performing channel estimation on an interfered pilot signal 601 to obtain an interfered channel sample 602 is schematically shown before inputting the initial neural network.

[0234] Exemplarily, for example, the initial neural network may also include a channel estimation subnetwork. The interfered pilot signal may be input into the initial neural network, and the channel estimation subnetwork of the initial neural network performs channel estimation on the interfered pilot signal to obtain an interfered channel sample.

[0235] Exemplarily, the feature extraction subnetwork may include, for example, multiple convolutional network layers.

[0236] In operation S612 , interference suppression is performed on the interference channel feature 603 according to the interference suppression sub-network N2 to obtain a sample channel feature 604 after interference suppression.

[0237] Exemplarily, the interference suppression sub-network may include: a recurrent neural network (RNN) and a long short-term memory network (LSTM).

[0238] The communication interference suppression method of the disclosed embodiment processes signals with temporal characteristics, which are suitable for the memory characteristics of recurrent neural networks and long-short-term memory networks. RNNs are sensitive to short-term memory but are weak to long-term memory. LSTM is a variant of RNN. Any network layer of LSTM can selectively remember the output of the previous network layer, resulting in better performance in long-term memory scenarios.

[0239] In operation S613 , channel recovery is performed on the interference suppressed sample channel feature 604 according to the channel recovery subnetwork N3 to obtain a sample interference suppressed channel 605 .

[0240] Exemplarily, the channel recovery subnetwork may also include, for example, multiple convolutional network layers.

[0241] The sample interference suppression signal includes a sample interference suppression channel.

[0242] The initial neural network includes a feature extraction subnetwork N1, an interference suppression subnetwork N2, and a channel recovery subnetwork N3.

[0243] On the one hand, interference is related to channel estimation. On the other hand, there is the following situation: in the uplink communication direction, for example, transmitted from a user device to a network, signal interference is mainly channel interference. Therefore, the communication interference suppression method of the embodiment of the present disclosure can suppress channel interference. Specifically, the interference channel characteristics obtained by the feature extraction subnetwork can characterize the characteristics of the interfered channel. The interference suppression subnetwork can perform interference suppression based on the interference channel characteristics to obtain sample channel characteristics after interference suppression. The sample channel characteristics after interference suppression can be restored to a sample interference suppression channel by the channel recovery subnetwork. The sample interference suppression channel is interference suppressed, so channel interference suppression can be achieved.

[0244] Of course, in the downlink communication direction, for example, transmitted from the network to the user equipment, channel interference suppression can also be performed using the communication interference suppression method of the embodiment of the present disclosure.

[0245] According to another embodiment of the communication interference suppression method of the present disclosure, for example, the following embodiment can be used to implement a specific example of determining a feedback value based on a sample interference suppression signal and an interference-free signal sample: determining a feedback value based on a sample interference suppression channel and an interference-free channel sample.

[0246] The interference-free channel samples are obtained by performing channel estimation on the interference-free signal samples, and the interference-free signal samples include the interference-free channel samples.

[0247] Exemplarily, the feedback value may be determined according to a difference between samples of the interference-suppressed channel and samples of the interference-free channel.

[0248] For example, the following formula (1-7) can also be used as a loss function, which measures the difference between the sample interference suppression channel and the interference-free channel sample. When the difference between the sample interference suppression channel and the interference-free channel sample is minimized, the loss function converges. min(f(H I )-H0) (1-7)

[0249] f(*) still characterizes the processing of the neural network.

[0250] According to the communication interference suppression method of the embodiment of the present disclosure, when performing channel interference suppression, interference-free channel samples are used as labels. By determining feedback values ​​based on the sample interference suppression channels and the interference-free channel samples, the channel interference suppression performance of the current initial neural network can be accurately characterized. Subsequently, the parameters of the initial neural network can be accurately adjusted based on the feedback values, thereby obtaining a neural network with better channel interference suppression performance.

[0251] FIG7A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure.

[0252] As shown in FIG7A , according to another embodiment of the communication interference suppression method of the present disclosure, for example, the following embodiment can be used to implement processing of interfered signal samples using an initial neural network to obtain a specific example of a sample interference suppression signal.

[0253] In operation S711 , feature extraction is performed on the interfered channel sample 702 and the interfered data signal 703 according to the feature extraction sub-network N1 to obtain interference channel features 704 and interference data features 705 , respectively.

[0254] The interfered signal samples include an interfered pilot signal 701 and an interfered data signal 703 .

[0255] The interfered channel sample 702 is obtained by performing channel estimation on the interfered pilot signal 701 .

[0256] In the example of FIG7A , a specific example of performing channel estimation on an interfered pilot signal 701 to obtain an interfered channel sample 702 is schematically shown before inputting into the initial neural network.

[0257] Exemplarily, for example, the initial neural network may also include a channel estimation subnetwork. The interfered pilot signal may be input into the initial neural network, and the channel estimation subnetwork of the initial neural network performs channel estimation on the interfered pilot signal to obtain an interfered channel sample.

[0258] Exemplarily, the feature extraction subnetwork may include, for example, multiple convolutional network layers.

[0259] In operation S712 , an interference estimation result 707 is determined based on the interference channel characteristics 704 and the non-interference channel characteristics 706 .

[0260] Exemplarily, the interference-free channel feature may be obtained by performing channel estimation and feature extraction on the interference-free signal samples.

[0261] In operation S713 , the interference suppression sub-network N3 performs interference suppression on the interference data feature 705 according to the interference estimation result 707 to obtain an interference-suppressed sample data feature 708 .

[0262] Exemplarily, the interference suppression sub-network may include: a recurrent neural network (RNN) and a long short-term memory network (LSTM).

[0263] In operation S714 , data recovery is performed on the interference-suppressed sample data features 708 according to the data recovery sub-network N3 to obtain sample interference-suppressed data 709 .

[0264] Exemplarily, the channel recovery subnetwork may also include, for example, multiple convolutional network layers.

[0265] The sample interference suppression signal includes sample interference suppression data.

[0266] The initial neural network includes a feature extraction subnetwork N1, an interference suppression subnetwork N2, and a data recovery subnetwork N3.

[0267] It should be noted that, as shown in FIG7B , when a signal includes a pilot signal and a data signal, the interference to the signal actually indicates that both the pilot signal and the data signal are subject to the same interference. FIG7B schematically illustrates an example in which the pilot signal in each frame of the signal is a demodulation reference signal DMRS. According to the communication interference suppression method of an embodiment of the present disclosure, the interference distribution of the interfered signal can be estimated, and the interference can be balanced and suppressed to obtain an interference-suppressed data signal (in the training phase of the neural network, the interference-suppressed data signal is sample interference suppression data).

[0268] In the downlink communication direction, for example, from the network to the user equipment, it is usually used to transmit data, so more attention is paid to the accuracy of data transmission. The communication interference suppression method according to the embodiment of the present disclosure can be used for data interference suppression. Specifically, the interference channel characteristics and interference data characteristics obtained by the feature extraction subnetwork can respectively characterize the characteristics of the interfered channel and the interfered data. The interference estimation results obtained by the interference channel characteristics and the interference-free channel characteristics can actually characterize the channel interference. Since the data signal is transmitted through the channel, the interference to the data signal is actually consistent with the channel interference. According to the interference estimation result (i.e., channel interference), the interference data characteristics can be accurately and efficiently suppressed for communication interference.

[0269] According to another embodiment of the communication interference suppression method of the present disclosure, for example, the following embodiment can be used to implement a specific example of determining a feedback value based on a sample interference suppression signal and an interference-free signal sample: determining a feedback value based on sample interference suppression data and an interference-free data sample.

[0270] The interference-free signal samples include interference-free data samples.

[0271] For example, the feedback value may be determined based on a difference between the sample interference suppression data and the interference-free data sample.

[0272] Exemplarily, for example, the sample interference suppression data and the interference-free data samples may be converted into constellation points respectively, and the feedback value may be determined according to the distance between the two constellation points.

[0273] According to the communication interference suppression method of the disclosed embodiment, when performing data signal interference suppression, interference-free data samples are used as labels. The feedback value determined based on the sample interference suppression data and the interference-free data samples can accurately characterize the data interference suppression performance of the current initial neural network. Subsequently, the parameters of the initial neural network can be accurately adjusted based on the feedback value, thereby obtaining a neural network with even better data interference suppression performance.

[0274] FIG8A schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure.

[0275] As shown in FIG8A , according to another embodiment of the present disclosure, a method for suppressing communication interference can be implemented by using a neural network to suppress wireless communication interference on a signal, so as to obtain a specific example of an interference suppression signal: using a feature extraction subnetwork to extract the working channel H w Perform feature extraction to obtain channel features. Use the interference suppression sub-network to perform interference suppression on the channel features to obtain the interference suppressed channel features. Use the channel recovery sub-network to perform channel recovery on the interference suppressed channel features to obtain the interference suppressed channel (the interference suppressed channel is the interference suppressed H shown in FIG8A). w ).

[0276] The signal includes a pilot signal. The working channel is obtained by performing channel estimation on the pilot signal.

[0277] The neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a channel recovery subnetwork.

[0278] The interference suppression signal includes an interference suppression channel.

[0279] The details of neural network training have been described in detail above. The trained neural network can be used for communication interference suppression. The operation process executed by the neural network is the same as the operation performed by the initial neural network during the training process of the neural network, and also has the corresponding technical effects, which will not be repeated here.

[0280] FIG8A schematically shows a schematic diagram of a neural network applied to interference suppression in wireless communications. It can be understood that the neural network is trained based on the initial neural network. The {H0} training set shown in FIG8A is used as a training sample in the process of training the neural network. For relevant content about neural network training, please refer to the above text and will not be repeated here.

[0281] FIG8B schematically shows a schematic diagram of a communication interference suppression method according to yet another embodiment of the present disclosure.

[0282] As shown in FIG8B , according to another embodiment of the present disclosure, a method for suppressing communication interference can be implemented by using a neural network to suppress wireless communication interference on a signal, so as to obtain a specific example of an interference suppression signal. w The signal and data signals are extracted to obtain channel features and data features, respectively. Interference estimation is performed based on the channel features and interference-free channel features to obtain a channel interference estimation result. The interference suppression subnetwork performs interference suppression on the data features based on the channel interference estimation result to obtain interference-suppressed data features. The data recovery subnetwork performs data recovery on the interference-suppressed data features to obtain interference-suppressed data (the interference-suppressed data is the constellation points after interference suppression shown in FIG8B ).

[0283] The signal includes a pilot signal and a data signal, and the working channel is obtained by performing channel estimation on the pilot signal.

[0284] The neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a data recovery subnetwork.

[0285] The interference-free channel feature is obtained by performing channel estimation and feature extraction on the interference-free signal, where the interference-free signal comes from the second communication subject.

[0286] The interference suppression signal includes interference suppression data.

[0287] Figure 8B schematically shows a schematic diagram of a neural network applied to wireless communication interference suppression. It can be understood that the neural network is obtained by training based on the initial neural network. The {H0} training set shown in Figure 8B is used as a training sample in the process of training the neural network. The transmitted constellation points (the transmitted constellation points are data signals) are used as labels in the process of training the neural network. The LDPC decoding module is used to decode low-density parity check codes (Low Density Parity Check Codes, abbreviated as LDPC). Relevant content about neural network training can be found above and will not be repeated here.

[0288] FIG9 schematically shows a block diagram of a communication interference suppression device according to an embodiment of the present disclosure.

[0289] As shown in FIG9 , the communication interference suppression device 900 of the embodiment of the present disclosure includes, for example, a signal acquisition module 910 and an interference suppression module 920.

[0290] The signal acquisition module 910 is configured to acquire a signal from the second communication subject.

[0291] The interference suppression module 920 is configured to use a neural network to perform wireless communication interference suppression on the signal to obtain an interference suppression signal.

[0292] According to another embodiment of the present disclosure, the communication interference suppression device may further include: a training start module, configured to send a training mode start instruction to start neural network training, the training mode start instruction is also used to instruct the interference source, the first communication subject and the second communication subject to start wireless communication in a non-interference mode, wherein, in the non-interference mode, the second communication subject obtains a first signal that is not interfered with by the interference source.

[0293] According to another embodiment of the present disclosure, the communication interference suppression device may further include: a neural network training module, configured to train an initial neural network using interference-free signal samples and interfered signal samples to obtain a neural network, wherein the interference-free signal samples are obtained based on the first signal, and the interfered signal samples are obtained based on the second signal, and the second signal is a signal obtained by the second communication subject that is interfered with by the interference source.

[0294] According to another embodiment of the present disclosure, the communication interference suppression device may further include: a training end module, configured to send a training mode end instruction, wherein the training mode end is used to instruct the first communication subject to end the neural network training, and the training mode end instruction is also used to instruct the second communication subject to send a signal to the first communication subject.

[0295] It should be understood that the embodiments of the device part of the present disclosure are the same or similar to the embodiments of the method part of the present disclosure, and the technical problems solved and the technical effects achieved are also the same or similar, and the present disclosure will not elaborate on them here.

[0296] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0297] FIG10 shows a schematic block diagram of an example electronic device 1000 that can be used to implement an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0298] As shown in FIG10 , the device 1000 includes a computing unit 1001 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. Various programs and data required for the operation of the device 1000 can also be stored in the RAM 1003. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0299] Various components in device 1000 are connected to I / O interface 1005, including an input unit 1006, such as a keyboard, mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a magnetic disk, optical disk, etc.; and a communication unit 1009, such as a network card, modem, wireless communication transceiver, etc. The communication unit 1009 allows device 1000 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0300] The computing unit 1001 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as the communication interference suppression method. For example, in some embodiments, the communication interference suppression method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 1000 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into the RAM 1003 and executed by the computing unit 1001, one or more steps of the communication interference suppression method described above can be performed. Alternatively, in other embodiments, the computing unit 1001 can be configured to perform the communication interference suppression method in any other appropriate manner (e.g., by means of firmware).

[0301] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0302] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0303] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0304] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0305] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0306] Computer systems may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The client and server relationship arises through computer programs running on the respective computers and having a client-server relationship to each other.

[0307] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0308] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A communication interference suppression method, applied to a first communication subject, characterized in that: include: Acquiring a signal from a second communication subject; A neural network is used to perform wireless communication interference suppression on the signal to obtain an interference suppression signal.

2. The method according to claim 1, characterized in that Before acquiring the signal from the second communication subject, the method further includes: A training mode start instruction is sent to start the neural network training, and the training mode start instruction is used to instruct the interference source, the first communication subject, and the second communication subject to start wireless communication in a non-interference mode, wherein in the non-interference mode, the second communication subject obtains a first signal that is not interfered by the interference source.

3. The method according to claim 2, characterized in that Also includes: An initial neural network is trained using interference-free signal samples and interfered signal samples to obtain the neural network, wherein the interference-free signal samples are obtained based on the first signal, and the interfered signal samples are obtained based on the second signal, and the second signal is a signal obtained by the second communication subject and interfered by the interference source.

4. The method according to claim 3, characterized in that Also includes: Sending a training mode end instruction, wherein the training mode end instruction is used to instruct the first communication subject to end the neural network training, and the training mode end instruction is also used to instruct the second communication subject to send the signal to the first communication subject.

5. The method according to any one of claims 1 to 4, characterized in that in, The signal includes a pilot signal, the neural network includes a feature extraction subnetwork, an interference suppression subnetwork, and a channel recovery subnetwork; and using the neural network to suppress wireless communication interference on the signal to obtain an interference suppression signal includes: Extracting features from a working channel using the feature extraction subnetwork to obtain channel features, wherein the working channel is obtained by performing channel estimation on the pilot signal; Using the interference suppression subnetwork to perform interference suppression on the channel characteristics to obtain the interference suppressed channel characteristics; The channel recovery subnetwork is used to perform channel recovery on the interference suppressed channel characteristics to obtain an interference suppressed channel, wherein the interference suppressed signal includes the interference suppressed channel.

6. The method according to any one of claims 1 to 4, characterized in that in, The signal includes a pilot signal and a data signal, and the neural network includes a feature extraction subnetwork, an interference suppression subnetwork and a data recovery subnetwork; The using a neural network to perform wireless communication interference suppression on the signal to obtain an interference suppression signal includes: Using the feature extraction subnetwork to extract features from a working channel and a data signal to obtain channel features and data features respectively, wherein the working channel is obtained by performing channel estimation on the pilot signal; performing interference estimation according to the channel characteristics and the interference-free channel characteristics to obtain a channel interference estimation result, wherein the interference-free channel characteristics are obtained by performing channel estimation and feature extraction on the interference-free signal, and the interference-free signal comes from the second communication subject; Using the interference suppression subnetwork, performing interference suppression on the data feature according to the channel interference estimation result to obtain the interference suppressed data feature; The data recovery subnetwork is used to perform data recovery on the interference suppressed data features to obtain interference suppressed data, wherein the interference suppressed signal includes the interference suppressed data.

7. The method according to claim 1, characterized in that in, The neural network is trained using the following operations: Using the initial neural network to process the interfered signal samples to obtain the sample interference suppression signal; Determining a feedback value according to the sample interference suppression signal and the interference-free signal sample; The parameters of the initial neural network are adjusted according to the feedback value to obtain the neural network.

8. The method according to claim 7, characterized in that in, The interfered signal samples include interfered pilot signals, and the using of the initial neural network to process the interfered signal samples to obtain a sample interference suppression signal includes: Extracting features from the interfered channel samples according to the feature extraction subnetwork to obtain interference channel features, wherein the interfered channel samples are obtained by performing channel estimation on the interfered pilot signal; Performing interference suppression on the interference channel characteristics according to the interference suppression subnetwork to obtain sample channel characteristics after interference suppression; Channel recovery is performed on the sample channel features after interference suppression according to the channel recovery subnetwork to obtain a sample interference suppressed channel, wherein the sample interference suppressed signal includes the sample interference suppressed channel, and the initial neural network includes the feature extraction subnetwork, the interference suppression subnetwork and the channel recovery subnetwork.

9. The method according to claim 8, characterized in that in, The determining of the feedback value according to the sample interference suppression signal and the loss function comprises: The feedback value is determined according to the sample interference suppression channel and the non-interference channel sample, wherein the non-interference channel sample is obtained by performing channel estimation on the non-interference signal sample, and the non-interference signal sample includes the non-interference channel sample.

10. The method according to claim 7, characterized in that in, The interfered signal samples include interfered pilot signals and interfered data signals, and the using of the initial neural network to process the interfered signal samples to obtain the sample interference suppression signal includes: Performing feature extraction on the interfered channel sample and the interfered data signal according to the feature extraction subnetwork to obtain interference channel features and interference data features respectively, wherein the interfered channel sample is obtained by performing channel estimation on the interfered pilot signal; Determining an interference estimation result according to the interference channel characteristics and the non-interference channel characteristics; The interference suppression subnetwork performs interference suppression on the interference data feature according to the interference estimation result to obtain the sample data feature after interference suppression; The sample data features after interference suppression are restored according to the data recovery subnetwork to obtain sample interference suppression data, wherein the sample interference suppression signal includes the sample interference suppression data, and the initial neural network includes the feature extraction subnetwork, the interference suppression subnetwork and the data recovery subnetwork.

11. The method according to claim 10, characterized in that in, The determining of the feedback value according to the sample interference suppression signal and the loss function comprises: The feedback value is determined according to the sample interference suppression data and the interference-free data sample, wherein the interference-free signal sample includes the interference-free data sample.

12. The method according to claim 7, characterized in that in, The initial neural network is a pre-trained neural network.

13. The method according to claim 8 or 10, characterized in that: in, The interference suppression subnetwork includes: a long short-term memory network.

14. A communication interference suppression device, applied to a first communication subject, characterized in that: include: A signal acquisition module, configured to acquire a signal from a second communication subject; The interference suppression module is configured to use a neural network to perform wireless communication interference suppression on the signal to obtain an interference suppression signal.

15. The device according to claim 14, characterized in that Also includes: A training start module is configured to send a training mode start instruction to start the neural network training, and the training mode start instruction is also used to instruct the interference source, the first communication subject, and the second communication subject to start wireless communication in a non-interference mode, wherein in the non-interference mode, the second communication subject obtains a first signal that is not interfered by the interference source.

16. The apparatus according to claim 15, further comprising: The neural network training module is configured to train an initial neural network using interference-free signal samples and interfered signal samples to obtain the neural network, wherein the interference-free signal samples are obtained based on the first signal, and the interfered signal samples are obtained based on a second signal, and the second signal is a signal obtained by the second communication subject and interfered by the interference source.

17. The device according to claim 16, characterized in that Also includes: The training end module is configured to send a training mode end instruction, wherein the training mode end instruction is used to instruct the first communication subject to end the neural network training, and the training mode end instruction is also used to instruct the second communication subject to send the signal to the first communication subject.

18. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the method according to any one of claims 1 to 13 is performed.

19. A non-transitory computer-readable storage medium storing computer instructions, wherein: When the computer instructions are executed, the method according to any one of claims 1 to 13 is performed.

20. A computer program product, comprising a computer program, wherein the computer program is stored in at least one of a readable storage medium and an electronic device, and when the computer program is executed by a processor, the method according to any one of claims 1 to 13 is performed.

Citation Information

Patent Citations

  • Electromagnetic interference noise suppression method and device, equipment and storage medium

    CN114510164A

  • Methods and apparatus for analog canceler tuning using neural networks

    US20230081695A1

  • Self-Interference Correction

    US20230119415A1

  • Terminal and base station

    WO2021000264A1

Cited By

  • Base station signal recovery method for realizing interference suppression based on deep neural network

    CN120692582A