KR20250100449A - Method and apparatus for estimating channel of multi antenna - Google Patents

Abstract

The following disclosure relates to a method for estimating a communication channel of a multi-antenna, which may include an operation of receiving received signals including input signals and non-orthogonal pilot signals received from one or more devices, an operation of estimating a channel of each of the devices using a learning-based channel estimation model based on the non-orthogonal pilot signals, an operation of decoding the input signals and outputting detection signals based on output values of the learning-based channel estimation model, and an operation of updating the learning-based channel estimation model using a predetermined loss function, the output values, the detection signals, and the input signals.

Description

{METHOD AND APPARATUS FOR ESTIMATING CHANNEL OF MULTI ANTENNA}

The following embodiments relate to a method and device for estimating a communication channel of multiple antennas.

Massive multiple-input and multiple-output (Massive MIMO) systems can meet the increasing user capacity and diverse requirements due to Massive Machine Type Communications (mMTC) and Ultra-Reliable Low Latency Communications (URLLC). Recently, Massive MIMO systems using Grant-Free (GF) multiple access have been studied to effectively satisfy the uplink transmission requirements of URLLC. Approaches to channel estimation in MIMO systems require orthogonal pilot signals, which may require synchronization between devices. This limitation may make it difficult to integrate GF multiple access with Massive MIMO, and thus, it may be necessary to use non-orthogonal pilots for channel estimation.

A method for estimating a communication channel of a multi-antenna according to one embodiment may include an operation of receiving received signals including input signals and non-orthogonal pilot signals received from one or more devices, an operation of estimating a channel of each of the devices using a learning-based channel estimation model based on the non-orthogonal pilot signals, an operation of decoding the input signals and outputting detection signals based on output values of the learning-based channel estimation model, and an operation of updating the learning-based channel estimation model using a predetermined loss function, the output values, the detection signals, and the input signals.

The above learning-based channel estimation model may be a model that estimates information on channels of the devices by relying on pilot information, and is learned using a semi-supervised learning method.

The above learning-based channel estimation model can learn to improve the accuracy of the channel estimation by learning the correlation between the channels.

The above-determined loss function may be a function that applies MSE (Mean Square Error) between the symbol value of the detected signal and the symbol value of the input signal using the information vector of the estimated channel included in the output value.

The operation of updating the learning-based channel estimation model may include an operation of executing the learning-based channel estimation model as many times as the number of multiple antennas, and then updating the learning-based channel estimation model.

The above receiving operation may include an operation of dividing the receive signal into a real part and an imaginary part.

The above input signal may include symbol modulation by OFDM (Orthogonal Frequency-Division Multiplexing).

The above non-orthogonal pilot signal can be generated using a random QPSK (Quadrature Phase Shift Keying) symbol.

The above decoding operation may include an operation of inputting the input signal and the output value of the learning-based channel estimation model to a matching filter.

An electronic device driving a Massive MIMO system according to one embodiment may include an antenna for receiving an input signal from one or more devices, a memory for storing instructions, and a processor, wherein the instructions, when executed by the processor, cause the electronic device to receive received signals including input signals and non-orthogonal pilot signals received from one or more devices, estimate a channel of each of the devices using a learning-based channel estimation model based on the non-orthogonal pilot signals, decode the input signal based on an output value of the learning-based channel estimation model to output a detection signal, and update the learning-based channel estimation model using a predetermined loss function, the output value, the detection signal, and the input signal.

The above learning-based channel estimation model may be a model that estimates the terminal's channel by relying on pilot information, and is learned using a semi-supervised learning method.

The above learning-based channel estimation model can learn to improve the accuracy of the channel estimation by learning the correlation between the channels.

The above-determined loss function may be a function that applies MSE (Mean Square Error) between the symbol value of the detected signal and the symbol value of the input signal using the information vector of the estimated channel included in the output value.

The above processor can update the learning-based channel estimation model after executing the learning-based channel estimation model as many times as the number of the multiple antennas.

The above processor can separate the receive signal into a real part and an imaginary part.

The above input signal may include symbol modulation by OFDM (Orthogonal Frequency-Division Multiplexing).

The above non-orthogonal pilot signal can be generated using a random QPSK (Quadrature Phase Shift Keying) symbol.

The above processor can input the input signal and the output value of the learning-based channel estimation model to the matching filter.

FIG. 1 is a flowchart illustrating a method for estimating a communication channel of multiple antennas according to one embodiment.
FIG. 2 is a diagram schematically illustrating an access process of a Massive MIMO system and device according to one embodiment.
FIG. 3 is a schematic diagram illustrating a Massive MIMO system according to one embodiment.
FIGS. 4 and 5 are schematic diagrams illustrating a learning-based estimation model according to one embodiment.
FIG. 6 is a block diagram illustrating an electronic device according to one embodiment.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be implemented in various forms. Accordingly, the actual implemented form is not limited to the specific embodiments disclosed, and the scope of the present disclosure includes modifications, equivalents, or alternatives included in the technical idea described in the embodiments.

Although the terms first or second may be used to describe various components, such terms should be construed only for the purpose of distinguishing one component from another. For example, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component.

When it is said that a component is "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but there may also be other components in between.

Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, the terms "comprises" or "has" and the like are intended to specify the presence of a described feature, number, step, operation, component, part, or combination thereof, but should be understood to not preclude the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

In this document, each of the phrases "A or B", "at least one of A and B", "at least one of A or B", "A, B, or C", "at least one of A, B, and C", and "at least one of A, B, or C" can include any one of the items listed together in that phrase, or all possible combinations of them.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning they have in the context of the relevant art, and will not be interpreted in an idealized or overly formal sense unless explicitly defined herein.

Hereinafter, embodiments will be described in detail with reference to the attached drawings. In describing with reference to the attached drawings, identical components are given the same reference numerals regardless of the drawing numbers, and redundant descriptions thereof will be omitted.

Massive MIMO systems can be important in the 5G physical layer because they can efficiently multiplex many devices using the same time-frequency resources. Massive MIMO systems can meet the diverse requirements introduced by mMTC and URLLC, as well as the requirement for larger user capacity.

In order to enable GF multiple access in Massive MIMO systems, pilot design for channel estimation of GF devices may be required. In conventional communications, orthogonal pilots may be used for channel estimation. On the other hand, in Massive MIMO systems for mMTC, pilot orthogonality problems may occur due to non-orthogonal pilots caused by pilot overhead. In particular, the probability of non-orthogonal pilots appearing may increase in GF environments.

In wireless communication, if the channel is maintained stably within a certain range, a longer packet length can reduce resource allocation for channel estimation, which can enable more efficient communication. However, a longer packet length can increase transmission latency when a packet is generated for a device that is not currently communicating. To solve the latency problem, GF multiple access can be utilized for transmission in a Massive MIMO system. Due to the nature of Massive MIMO, when the data traffic is not very high, even when a GB (Grant-based) device transmits a packet, the base station can accommodate additional GF devices.

FIG. 1 is a flowchart illustrating a method for estimating a communication channel of multiple antennas according to one embodiment.

For convenience of explanation, operations (110 to 140) are described as being performed using the electronic device (600) illustrated in FIG. 6. However, these operations (110 to 140) may be utilized via any other suitable electronic device and within any suitable system.

Furthermore, the operations of FIG. 1 may be performed in the order and manner illustrated, but the order of some operations may be changed or some operations may be omitted without departing from the spirit and scope of the illustrated embodiment. A plurality of operations illustrated in FIG. 1 may be performed in parallel or simultaneously. Hereinafter, the electronic device (600) may be referred to as a base station (BS) (e.g., the base station (300) of FIG. 3).

In operation (110), the electronic device (600) may receive an input signal (e.g., an input signal (302) of FIG. 3) and a receive signal (e.g., a receive signal (303) of FIG. 3) including a non-orthogonal pilot signal (e.g., a non-orthogonal pilot signal (301) of FIG. 3) from one or more devices (e.g., devices (310) of FIG. 3). The device may refer to an electronic device capable of transmitting and receiving data or signals with a base station. The device may refer to a sensor of the Internet of Things (IoT), a smartphone, a self-driving car, and the like. In the description below, the device may include both a GB device and a GF device, but a device that communicates with a Massive MIMO base station using a non-orthogonal pilot signal may refer to a GF device.

The electronic device (600) is a base station that operates a Massive MIMO system and may include M antennas (e.g., antennas (630) of FIG. 6). The base station may perform uplink communication with N devices. In the described embodiment, the device is assumed to have a single antenna, but is not limited thereto and may include multiple antennas.

It is assumed that the device and the BS communicate over a channel following Rayleigh fading. Rayleigh fading can be a type of signal fading that occurs in a wireless communication environment. Rayleigh fading can be used to model rapid changes in signal strength over a short period of time or over a travel distance. Rayleigh fading can be used in environments where there are many obstacles and no clear line-of-sight path between the transmitter and the receiver. In Rayleigh fading, the signal amplitude change can be modeled as following the Rayleigh distribution, which is a statistical model for random variables. It is also assumed that there is Additive White Gaussian Noise (AWGN) in the communication between the device and the BS.

AWGN can be a basic noise model used in information theory to mimic the effects of many random processes occurring in nature. AWGN can be characterized by having the same intensity at different frequencies in a white spectrum and producing a constant power spectral density. AWGN can have a 'Gaussian' distribution, as it has a normal distribution in the time domain with a zero mean and a constant standard deviation. The AWGN model can be used in communication system simulations to represent the effects of random noise on signal transmission.

An input signal received at the antennas of a base station according to one embodiment can be expressed as a signal of a t -th OFDM (Othogonal Frequency-Division Multiplexing) symbol, as shown in the following mathematical expression 1.

In mathematical expression 1, may mean a channel between the i -th GF device and the BS. may mean the t-th OFDM symbol of the i-th GF device. can mean zero-mean iid (independent and identically distributed) AWGN of the t-th OFDM symbol.

The input signal may include symbol modulation by OFDM. Orthogonal Frequency-Division Multiplexing (OFDM) may be a method of encoding digital data on multiple carrier frequencies. OFDM can be used in modern wideband communication systems such as Wi-Fi, LTE, and 5G because of its efficiency in dealing with problems such as multipath fading and interference. OFDM works by splitting data across many subcarriers that are close together, each of which can be modulated at a low symbol rate. This design allows for robustness against channel imperfections and efficient use of spectrum, making OFDM an effective solution for high-speed data transmission in wireless networks.

In wireless communication, since the frame structure is defined, the length of the pilot may not be arbitrarily extended. For example, in the frame structure of LTE, one time slot consists of 7 OFDM symbols, one subframe consists of 2 time slots, and one frame can consist of 10 subframes. Accordingly, one frame corresponding to a 10-millisecond packet can consist of 140 OFDM symbols. Therefore, the number of OFDM symbols in one packet can be represented as Ns , and the number of pilots can be represented as Np( , p can be represented as an overhead value. In the following description, it is assumed that 140 OFDM symbols are transmitted, and 11 of them are designated as pilot symbols ( p = 11 ). However, it is not limited to the described embodiment, and the length of the frame may vary.

In addition, it is assumed that QPSK (Quadrature Phase Shift Keying) is used as the modulation method of the pilot symbol. Likewise, non-orthogonal pilot signals can be generated using random QPSK symbols.

QPSK can be a form of phase modulation used in digital communication systems. QPSK can represent data by changing the phase of a reference signal (e.g., a carrier). In QPSK, each symbol can represent two bits of data. The phase of the carrier can take one of four different values during each symbol period. These values can be 90 degrees apart (Quadrature). QPSK can achieve a balance between data rate and signal robustness.

In operation (120), the electronic device (600) may estimate a channel of each of the devices using a learning-based channel estimation model (e.g., an artificial neural network model (320) of FIG. 3) based on a non-orthogonal pilot signal. Here, estimating a channel by the electronic device (600) may mean estimating information of the channel. The information of the channel may mean a vector corresponding to a direction of the channel of each of the devices.

A learning-based channel estimation model can be a model that estimates information about the channels of devices by relying on pilot information, and is learned using a semi-supervised learning method.

Learning-based channel estimation models can estimate channel information using non-orthogonal pilots. Non-orthogonal pilot signals can use random QPSK symbols to enable asynchronous transmission between devices. Learning-based channel estimation models can use a semi-supervised learning approach that relies on pilot information for channel estimation.

Semi-supervised learning can be a machine learning approach that combines a small amount of labeled data with a large amount of unlabeled data during training. Semi-supervised learning methods can be considered a learning method that is somewhere between supervised learning (where all data is labeled) and unsupervised learning (where the data is unlabeled).

Semi-supervised learning models can learn the underlying structure and features of the data using labeled data, and then apply the learning results to a larger set of unlabeled data. This can be useful in cases where labeled data is expensive or time-consuming to obtain.

A learning-based channel estimation model may be a model that produces different output values depending on the number of devices, or may include multiple versions of the model that can be applied differently depending on the number of devices.

The learning-based channel estimation model can learn to improve the accuracy of channel estimation by learning the correlation between channels. The channels between devices may interfere with each other or may have correlations. Therefore, the learning-based channel estimation model can improve the accuracy of the channel estimation model by learning the correlation between channels during the learning process. Since the learning-based channel estimation model can be constructed using an artificial neural network model (e.g., a deep neural network (DNN)), it can also learn the correlation between channels.

In operation (130), the electronic device (600) can decode the receive signal and output a detection signal based on the output value of the learning-based channel estimation model. Here, the output value of the learning-based channel estimation model can mean information of the estimated channel, a vector corresponding to the direction of the estimated channel, or an information vector of the estimated channel.

An electronic device (600) according to one embodiment can input an input signal and an output value of a learning-based channel estimation model to a matched filter.

Matching filters are used in decoding in a Massive MIMO system, and can enable data decoding of a desired device if channel information of the desired device is known even when channel information of other devices is not available. For example, assume that a Massive MIMO system decodes the t -th OFDM symbol of the n- th device. In this case, the symbol can be decoded by the following method of mathematical expression 2.

In mathematical expression 2, can mean the vector of estimated channels of the n -th device. If M is a sufficiently large number, silver , it can converge to 0. As a result, mathematical expression 2 can be expressed as the following mathematical expression 3.

Therefore, the Massive MIMO system can decode the signal (or data) of the desired n -th device without channel information of other devices.

In operation (140), the electronic device (600) can update a learning-based channel estimation model using a predetermined loss function, an output value, a detection signal, and an input signal.

The predetermined loss function may be a function that applies the MSE (Mean Square Error) between the symbol value of the detected signal and the symbol value of the input signal using the information vector of the estimated channel included in the output value. The predetermined loss function can be expressed as in the following mathematical expression 4.

In mathematical expression 4, Np can be the length of the non-orthogonal pilot signal. can be a channel information vector of the i -th device. may be a received signal (e.g., input signal) for the t -th OFDM transmission symbol of the i -th device. The predetermined loss function may be a function that performs MSE on the values of the actual transmission symbol (e.g., signal transmitted by the device) and the detected symbol (e.g., output result detected by calculating the estimated channel information vector and the input signal). In order to apply the predetermined loss function, the output values of the learning-based channel estimation model may need to be collected as many as the number of antennas of the base station. Therefore, the learning-based channel estimation model may be updated once after execution as many as the number of antennas. The learning-based channel estimation model may be trained until the value of the predetermined loss function converges to a value close to 0. In the learning-based channel estimation model, the closer the value of the predetermined loss function is to 0, the more accurate the detected symbol value may be. Therefore, the more accurate the detected symbol value is, the The vector is a real channel vector And it can have a nice direction.

The input signals received from each antenna of the base station and the corresponding non-orthogonal pilots can be used as inputs for the learning-based channel estimation model. If the number of antennas is M, the learning-based channel estimation model can be executed M times to obtain channel information corresponding to the M antennas. After executing the learning-based channel estimation model M times, the output vectors are collected. You can get a vector. Vectors can be vectors with similar direction to the channel for the N terminals. Knowing the direction of the channel vector, the transmission signal of the device can be detected using the decoding method of the matching filter.

An electronic device (600) according to one embodiment can update a learning-based channel estimation model after executing the learning-based channel estimation model as many times as the number of multiple antennas. Since the learning-based channel estimation model does not depend on data transmitted from devices, learning is possible and can be updated even during continuous communication.

FIG. 1 is a diagram schematically illustrating an access process of a Massive MIMO system and a device according to one embodiment.

The description referring to Fig. 1 can be equally applied to Fig. 2, and any duplicate content can be omitted.

Referring to FIG. 2, a Massive MIMO base station (e.g., base station (300) of FIG. 3) can communicate with GB devices and GF devices. A GB device that does not require low-latency communication can communicate with a BS having a Massive MIMO system through a communication channel connection based on an orthogonal pilot. In the description below, it is assumed that interference by the GB device is removed by the Successive Interference Cancellation (SIC) method. A conventional Massive MIMO system cannot estimate the channel of a GF device, which may result in the cancellation of transmission packets of the GB device and the GF device. However, due to the channel hardening characteristic of Massive MIMO, a matched filter can be used to decode based on information about the channel of the GB device. A Massive MIMO system can enable decoding using channel hardening when there are not many devices.

FIG. 3 is a schematic diagram for explaining a system of a Massive MIMO base station according to one embodiment.

The description referring to FIG. 1 and FIG. 2 can be equally applied to FIG. 3, and any duplicate content can be omitted.

Referring to FIG. 3, one or more devices (310) may communicate with a Massive MIMO base station (300) (e.g., an electronic device (600) of FIG. 6) having multiple antennas (e.g., an antenna (630) of FIG. 6) to transmit an input signal (302). Here, a pilot (e.g., a non-orthogonal pilot (301)) may be inserted into the input signal. A random QPSK symbol may be used as a pilot signal to enable pilot-based channel estimation without pilot synchronization. A receive signal (303) received from each antenna of the base station (300) may be input to an artificial neural network model (320) (e.g., a learning-based channel estimation model) to obtain channel information. The base station (300) may utilize the obtained channel information to extract a detection signal (304).

FIGS. 4 and 5 are schematic diagrams illustrating a learning-based estimation model according to one embodiment.

The description referring to FIGS. 1 to 3 can be equally applied to FIGS. 4 and 5, and any overlapping content can be omitted.

Referring to FIGS. 4 and 5, the learning-based channel estimation model (320) may have a connected network structure composed of four hidden layers. The signal input to the learning-based channel estimation model (320) may be a complex value, while the weight of the neural network may be a real value. Therefore, it is necessary to separate the receive signals (303-1 to 303- N r) input to the learning-based channel estimation model (320) into a real part and an imaginary part. That is, instead of using the entire input signal (302) as input data, the Massive MIMO system may divide the signals received from each antenna (630-1 to 630- N r) (e.g., antenna (630) of FIG. 6) and the corresponding pilot into a real part and an imaginary part and utilize them as inputs to the learning-based estimation model (320). This method can prevent the size of the input signal (302) from increasing in proportion to the number of received antennas.

The output data of the learning-based channel estimation model (320) may mean channel information of each individual antenna. The Massive MIMO system may obtain channel information vectors (322) of each of the devices (310) based on the receive signal (303) received from each of the multiple antennas (630) and the output value (321) of the learning-based channel estimation model (320). In addition, the base station (300) may update the learning-based channel estimation model (320) using a predetermined loss function (323).

FIG. 6 is a block diagram illustrating an electronic device according to one embodiment.

One or more of the blocks and combinations of blocks of FIG. 6 may be implemented by a special purpose hardware-based computer performing a specific function, or a combination of special purpose hardware and computer instructions. The contents described with reference to FIGS. 1 to 5 may be equally applied to FIG. 6. For example, an electronic device (600) according to one embodiment may include a base station (300).

As shown in FIG. 6, the electronic device (600) may include a memory (610), a processor (620), and an antenna (630). The electronic device (600) may further include a communication module, and the communication module may include a transmitter and a receiver. In addition, the communication module may be connected to the antenna (630) to receive input signals (or data) from devices (e.g., devices (310) of FIG. 3).

An electronic device (600) according to one embodiment may include a memory (610) and a processor (620) connected to the memory (610) via a system bus or other suitable circuitry.

The electronic device (600) may store program code in memory (610). The memory (610) according to one embodiment may include one or more physical memory devices, such as local memory or one or more bulk storage devices. In this case, the local memory may include RAM (Random Access Memory) or other volatile memory devices that are generally used while actually executing the program code. The bulk storage device may be implemented as a Hard Disk Drive (HDD), a Solid State Drive (SSD), or other non-volatile memory device.

As the executable program code stored in the memory (610) is executed by the electronic device (600), various operations described in the present disclosure can be performed by the processor (620). For example, the memory (610) can store program code for causing the processor (620) to perform one or more operations described in FIGS. 1 to 5.

Depending on the particular type of device being implemented, the electronic device (600) may include fewer components than those illustrated or additional components not illustrated in FIG. 6. Additionally, one or more of the components may be incorporated into, or otherwise form part of, another component.

The processor (620) according to one embodiment is a hardware configuration that performs overall control functions for controlling operations of the electronic device (600). For example, the processor (620) may control the electronic device (600) overall by executing programs stored in the memory (610) within the electronic device (600). The processor (620) may be implemented as a central processing unit (CPU), a graphics processing unit (GPU), an application processor (AP), a neural processing unit (NPU), etc., provided within the electronic device (600), but is not limited thereto.

The processor (620) may cause the electronic device (600) to receive, when the instructions are executed, received signals including input signals and non-orthogonal pilot signals received from one or more devices, estimate a channel of each of the devices using a learning-based channel estimation model based on the non-orthogonal pilot signals, decode the input signals based on an output value of the learning-based channel estimation model to output a detection signal, and update the learning-based channel estimation model using a predetermined loss function, the output value, the detection signal, and the input signal.

The embodiments described above may be implemented as hardware components, software components, and/or a combination of hardware components and software components. For example, the devices, methods, and components described in the embodiments may be implemented using a general-purpose computer or a special-purpose computer, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions (or instructions). The processing device may execute an operating system (OS) and software applications running on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing device is sometimes described as being used alone, but those skilled in the art will appreciate that the processing device may include multiple processing elements and/or multiple types of processing elements. For example, a processing device may include multiple processors, or a processor and a controller. Other processing configurations, such as parallel processors, are also possible.

The software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing device to perform a desired operation or may independently or collectively command the processing device. The software and/or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal waves, for interpretation by the processing device or for providing instructions or data to the processing device. The software may also be distributed over network-connected computer systems and stored or executed in a distributed manner. The software and data may be stored on a computer-readable recording medium.

The method according to the embodiment may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., alone or in combination, and the program commands recorded on the medium may be those specially designed and configured for the embodiment or may be those known to and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program commands such as ROMs, RAMs, flash memories, etc. Examples of program commands include not only machine language codes generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, etc.

The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Although the embodiments have been described with limited drawings as described above, those skilled in the art can apply various technical modifications and variations based on them. For example, even if the described techniques are performed in a different order than the described method, and/or the components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or are replaced or substituted by other components or equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents to the claims are also included in the scope of the claims described below.

Claims (19)

In a method for estimating a communication channel of multiple antennas,
An operation of receiving received signals including input signals and non-orthogonal pilot signals received from one or more devices;
An operation of estimating a channel of each of the devices using a learning-based channel estimation model based on the non-orthogonal pilot signal;
An operation of decoding the input signal and outputting a detection signal based on the output value of the learning-based channel estimation model; and
An operation of updating the learning-based channel estimation model using a predetermined loss function, the output value, the detection signal, and the input signal.
A method for estimating a communication channel of multiple antennas, comprising:
In the first paragraph,
The above learning-based channel estimation model
A multi-antenna communication channel estimation method, which is a model that estimates information about the channel of the devices based on pilot information by learning using a semi-supervised learning method.
In the first paragraph,
The above learning-based channel estimation model
A multi-antenna communication channel estimation method that learns correlations between the channels to improve the accuracy of channel estimation.
In the first paragraph,
The above predetermined loss function is
A multi-antenna communication channel estimation method, which is a function that applies MSE (Mean Square Error) between the symbol value of a signal detected using the information vector of the estimated channel included in the output value and the symbol value of the input signal.
In the first paragraph,
The action of updating the above learning-based channel estimation model is
An operation of updating the learning-based channel estimation model after executing the learning-based channel estimation model as many times as the number of the multiple antennas.
A method for estimating a multi-antenna communication channel, comprising:
In the first paragraph,
The above receiving action is
The operation of dividing the above received signal into real and imaginary parts.
A method for estimating a communication channel of multiple antennas, comprising:
In the first paragraph,
The above input signal is
A method for estimating a communication channel using multiple antennas, including symbol modulation using OFDM (Orthogonal Frequency-Division Multiplexing).
In the first paragraph,
The above non-orthogonal pilot signals are
A method for estimating a multi-antenna communication channel generated using random QPSK (Quadrature Phase Shift Keying) symbols.
In the first paragraph,
The above decoding operation is
An operation of inputting the input signal and the output value of the learning-based channel estimation model to the matching filter.
A method for estimating a multi-antenna communication channel, comprising:
A computer program stored on a computer-readable recording medium for executing the method of any one of claims 1 to 9 in combination with hardware.
In an electronic device driving a Massive MIMO system,
An antenna for receiving input signals from one or more devices;
Memory for storing instructions; and
Processor
Including,
The above instructions, when executed by the processor, cause the electronic device to:
Receives received signals including input signals and non-orthogonal pilot signals received from one or more devices,
Based on the non-orthogonal pilot signals, the channel of each of the devices is estimated using a learning-based channel estimation model,
Based on the output value of the above learning-based channel estimation model, the input signal is decoded to output a detection signal,
An electronic device configured to update the learning-based channel estimation model using a predetermined loss function, the estimated channel, the output value, and the input signal.
In Article 11,
The above learning-based channel estimation model
An electronic device, which is a model that estimates information of a channel of the devices based on pilot information and is learned using a semi-supervised learning method.
In Article 11,
The above learning-based channel estimation model
An electronic device that learns correlations between the channels to improve the accuracy of channel estimation.
In Article 11,
The above predetermined loss function is
An electronic device, which is a function that applies MSE (Mean Square Error) between the symbol value of a signal detected using the information vector of the estimated channel included in the output value and the symbol value of the input signal.
In Article 11,
The above processor
An electronic device that updates the learning-based channel estimation model after executing the learning-based channel estimation model as many times as the number of the multiple antennas.
In Article 11,
The above processor
An electronic device that divides the received signal into a real part and an imaginary part.
In Article 11,
The above input signal is
An electronic device including symbol modulation by OFDM (Orthogonal Frequency-Division Multiplexing).
In Article 11,
The above non-orthogonal pilot signals are
An electronic device that generates signals using random QPSK (Quadrature Phase Shift Keying) symbols.
In Article 11,
The above processor
An electronic device that inputs the input signal and the output value of the learning-based channel estimation model to the matching filter.

Images