Intelligent cascade channel feedback method for smart metasurface wireless communication

Abstract

This invention discloses a method for intelligent cascade channel feedback for intelligent metasurface wireless communications, belonging to the field of communications technology. In this method, during the first feedback within the base station-intelligent metasurface channel coherence time, the user compresses the base station-intelligent metasurface-user cascade channel and feeds it back to the base station. In subsequent channel feedback, the user calculates the ratio of the current base station-each intelligent metasurface unit-user cascade channel to the base station-each intelligent metasurface unit-user cascade channel in the initial feedback, compresses the vector consisting of these ratios, and feeds it back to the base station. This method significantly reduces the feedback overhead and complexity of intelligent metasurface-assisted communication systems.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an intelligent cascade channel feedback method for intelligent metasurface wireless communication. Background Art

[0002] Reconfigurable Intelligent Surface (RIS) is a potential key technology in future mobile communication systems. RIS consists of multiple controllable units. By regulating the electromagnetic response of these units, the propagation environment of wireless signals can be changed. The efficient regulation of RIS depends on the acquisition of downlink channel information. In a frequency division duplex (FDD) system, because the uplink and downlink channels operate in different frequency bands and are not mutually exclusive, the user equipment (UE) needs to feed back the estimated downlink CSI to the base station (BS). This process consumes a large amount of uplink resources. How to efficiently feed back downlink channel information to the BS is a key issue in wireless communications using intelligent metasurfaces.

[0003] RIS, with its low cost, has become a hot topic in the RIS field. In RIS wireless communications, since the controllable units lack signal transmission and reception capabilities, it is difficult for the UE to estimate the BS-RIS channel and the RIS-UE channel separately. Often, the UE can only estimate the BS-RIS-UE cascade channel, which is composed of the BS-each RIS unit-UE channel. In RIS-assisted FDD wireless communication systems, the UE needs to feed back the BS-RIS-UE cascade channel to the base station. Since the RIS is often located in a fixed position, the BS-RIS channel has a slow time-varying characteristic. Utilizing this characteristic is one of the key means of reducing the channel feedback overhead of RIS wireless communication systems. However, in RIS-assisted wireless communication systems, since the BS-RIS channel and the RIS-UE channel cannot be estimated separately, how to effectively utilize the slow time-varying characteristics of the BS-RIS channel to reduce feedback overhead has become a major challenge in RIS applications. Summary of the Invention

[0004] The present invention provides an intelligent cascade channel feedback method for intelligent metasurface wireless communication, which can effectively utilize the slow time-varying characteristics of the BS-RIS channel to reduce channel feedback overhead and computational complexity when the UE cannot estimate the BS-RIS channel and the RIS-UE channel respectively.

[0005] An embodiment of the present invention provides an intelligent cascade channel feedback method for intelligent metasurface wireless communication, comprising the following steps: Step 1, constructing two autoencoder neural networks, including autoencoder 1 and autoencoder 2, wherein autoencoder 1 is composed of encoder 1 and decoder 1, and is used to feedback the base station-intelligent metasurface-user cascade channel, and autoencoder 2 is composed of encoder 2 and decoder 2, and is used to feedback the ratio vector composed of the ratio of the base station-each intelligent metasurface unit-user cascade channel at adjacent moments;

[0006] Step 2: During the coherence time of the base station-intelligent metasurface channel, when the channel feedback is performed for the first time, the user uses encoder 1 to compress the cascade channel H(0) into a feedback bit stream s1(0) and transmits it to the base station. After receiving the feedback bit stream s1(0), the base station uses decoder 1 to reconstruct the cascade channel.

[0007] Step 3: During the second and subsequent channel feedback within the coherence time of the base station-intelligent metasurface channel, the user calculates the ratio between the base station-each intelligent metasurface unit-user cascade channel at the current moment and the base station-each intelligent metasurface unit-user cascade channel of the first feedback, forming a proportional vector p(t). The proportional vector p(t) is then compressed into a feedback bit stream s2(t) using encoder 2 and transmitted to the base station. After receiving the feedback bit stream s2(t), the base station reconstructs the proportional vector using decoder 2. And based on the cascade channel reconstructed from the first channel feedback With the proportional vector Calculate the reconstructed cascade channel at the current moment

[0008] Optionally, in one embodiment of the present invention, the user is configured with a single antenna, and the cascade channel H(t) is composed of the product of the diagonal matrix consisting of the base station-intelligent metasurface channel matrix B(t) and the intelligent metasurface-user channel vector a(t), which is described as follows:

[0009] H(t)=diag(a(t))B(t)

[0010] Where diag(·) is the transformation that converts a vector into a diagonal matrix.

[0011] Optionally, in one embodiment of the present invention, the base station-i-th intelligent metasurface unit-user cascade channel is the i-th row h of the cascade channel H(t) matrix. i (t), the i-th element p of the ratio vector p(t) i The calculation method of (t) is described as follows:

[0012]

[0013] Here, the superscript H denotes the conjugate transpose and ‖·‖2 is the Euclidean norm.

[0014] Optionally, in one embodiment of the present invention, the cascade channel reconstructed from the first channel feedback is With the proportional vector Calculate the reconstructed cascade channel at the current moment The method is described as follows:

[0015]

[0016] in, To reconstruct the scale vector The i-th element in .

[0017] Optionally, in one embodiment of the present invention, encoder 1 is composed of a neural network, which is used to compress the base station-intelligent metasurface-user cascade channel H(0), and the cascade channel H(0) is the input of encoder 1. The output of the neural network of encoder 1 is quantized to form a feedback bit stream s1(0), which is transmitted to the base station through the uplink control channel.

[0018] Optionally, in one embodiment of the present invention, encoder 2 is composed of a neural network, which is used to compress the proportional vector p(t), and the proportional vector p(t) is the input of encoder 2. The output of the neural network of encoder 2 is quantized to form a feedback bit stream s2(t), which is transmitted to the base station through the uplink control channel.

[0019] Optionally, in one embodiment of the present invention, the decoder 1 is composed of a neural network and is used to reconstruct the feedback cascade channel through the feedback bit stream s1(0). The input of the decoder 1 is the feedback bit stream s1(0), and the output is the reconstructed cascade channel.

[0020] Optionally, in one embodiment of the present invention, the decoder 2 is composed of a neural network and is used to reconstruct the feedback cascade channel through the feedback bit stream s2(t). The input of the decoder 2 is the feedback bit stream s2(t), and the output is the reconstructed scale vector

[0021] Optionally, in one embodiment of the present invention, the autoencoder 1 uses the input cascade channel H(t) and the reconstructed cascade channel The mean square error between is used as the cost function for training to minimize the cost function. During the training process, the gradient is set to a constant 1. The cost function is described as follows:

[0022]

[0023] where ‖·‖2 is the Euclidean norm.

[0024] Optionally, in one embodiment of the present invention, the autoencoder 2 uses the input scale vector p(t) and the reconstructed scale vector The mean square error between is used as the cost function for training to minimize the cost function. During the training process, the gradient is set to a constant 1. The cost function is described as follows:

[0025]

[0026] where ‖·‖2 is the Euclidean norm.

[0027] The intelligent cascade channel feedback method for intelligent metasurface wireless communication in an embodiment of the present invention takes into account the defects in intelligent metasurface wireless communication. When the base station-intelligent metasurface channel and the intelligent metasurface-user channel cannot be estimated and only the cascade channel can be estimated, the method effectively utilizes the slow time-varying characteristics of the base station-intelligent metasurface channel to reduce the feedback overhead and reduce the computational complexity of the feedback algorithm, thereby realizing low feedback overhead and low complexity channel state information in intelligent metasurface wireless communication.

[0028] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0030] Figure 1 A flowchart of an intelligent cascade channel feedback method for intelligent metasurface wireless communication provided according to an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a simulation scenario according to an embodiment of the present invention;

[0032] Figure 3 Schematic diagram of the transmission interval of the concatenated channel H(t) and the proportional vector p(t) according to an embodiment of the present invention;

[0033] Figure 4 It is the overall implementation framework of the embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0035] Figure 1 This is a flowchart of an intelligent cascade channel feedback method for intelligent metasurface wireless communication provided according to an embodiment of the present invention.

[0036] like Figure 1 As shown, the intelligent cascade channel feedback method for intelligent metasurface wireless communication includes the following steps:

[0037] Step 1, construct two autoencoder neural networks, including autoencoder 1 and autoencoder 2, wherein autoencoder 1 consists of encoder 1 and decoder 1, and is used to feedback the base station-intelligent metasurface-user cascade channel (i.e., channel H(0)), and autoencoder 2 consists of encoder 2 and decoder 2, and is used to feedback the ratio vector composed of the ratio of the base station-each intelligent metasurface unit-user cascade channel at adjacent times (i.e., vector p(t), where t represents the time).

[0038] Step 2: During the coherence time of the base station-intelligent metasurface channel, when the channel feedback is performed for the first time, the user uses encoder 1 to compress the cascade channel H(0) into a feedback bit stream s1(0) and transmits it to the base station. After receiving the feedback bit stream s1(0), the base station uses decoder 1 to reconstruct the cascade channel.

[0039] Step 3: During the second and subsequent channel feedback within the coherence time of the base station-intelligent metasurface channel, the user calculates the ratio between the base station-each intelligent metasurface unit-user cascade channel at the current moment and the base station-each intelligent metasurface unit-user cascade channel of the first feedback, forming a proportional vector p(t). The proportional vector p(t) is then compressed into a feedback bit stream s2(t) using encoder 2 and transmitted to the base station. After receiving the feedback bit stream s2(t), the base station reconstructs the proportional vector using decoder 2. And based on the cascade channel reconstructed from the first channel feedback With the proportional vector Calculate the reconstructed cascade channel at the current moment

[0040] The present invention considers a large-scale multi-input multi-output narrowband wireless communication system assisted by an intelligent metasurface, where the base station is equipped with 32 transmitting antennas, the user is equipped with a single receiving antenna, and the intelligent metasurface is equipped with 16×16 reflective units. Figure 2As shown in Figure 1, the signal transmission path between the base station and the user is completely blocked, and the only signal transmission path can be established through the smart metasurface. Here, the base station-smart metasurface channel is represented as B(t), with dimensions of 256×32, the smart metasurface-user channel is represented as a(t), with dimensions of 256×1, and the base station-smart metasurface-user cascade channel is represented as H, with dimensions of 32×256. Direct feedback of the base station-smart metasurface-user cascade channel H(t) will incur significant feedback overhead.

[0041] In the embodiment of the present invention, the user is configured with a single antenna, and the cascade channel H(t) is composed of the product of the diagonal matrix consisting of the base station-intelligent metasurface channel matrix B(t) and the intelligent metasurface-user channel vector a(t), which is described as follows:

[0042] H(t)=diag(a(t))B(t)

[0043] Where diag(·) is the transformation that converts a vector into a diagonal matrix.

[0044] The base station-i-th intelligent metasurface unit-user cascade channel is the i-th row h of the cascade channel matrix H(t) i (t), the i-th element p of the ratio vector p(t) i The calculation method of (t) is described as follows:

[0045]

[0046] Here, the superscript H denotes the conjugate transpose and ‖·‖2 is the Euclidean norm.

[0047] The cascade channel reconstructed from the first channel feedback With the proportional vector Calculate the reconstructed cascade channel at the current moment The method is described as follows:

[0048]

[0049] in, To reconstruct the scale vector The i-th element in .

[0050] In the embodiments of the present invention, the base station and the smart metasurface both have fixed geographical locations, and the environment between the base station and the smart metasurface is considered to be relatively fixed, and the base station-smart metasurface channel B(t) remains stable for a long period of time. However, the user has mobility, and the environment between the smart metasurface and the user is considered to be relatively non-fixed, and the smart metasurface-user channel changes faster.

[0051] Specifically, since the base station and the smart metasurface are both fixed and often located far from the ground, the propagation environment between them is relatively stable, and the base station-smart metasurface channel coherence time is long. However, due to the mobility of users, the environment between the smart metasurface and the user is constantly changing, and the smart metasurface-user channel coherence time is relatively short. Therefore, Figure 3 As shown, in this method, the base station-intelligent metasurface-user cascade channel carrying the base station-intelligent metasurface channel does not need to be frequently fed back, but the ratio vector carrying the information related to the smart metasurface-user channel needs to be frequently fed back.

[0052] The smart metasurface does not have the ability to transmit and receive wireless signals. It is believed that in smart metasurface wireless communications, the base station-smart metasurface channel B(t) and the smart metasurface-user channel a(t) cannot be obtained in channel estimation. Only the base station-smart metasurface-user channel H(t) can be obtained through channel estimation.

[0053] Figure 4 The overall architecture of the implemented method is presented. The neural network used includes autoencoder 1 and autoencoder 2. Autoencoder 1 consists of encoder 1 on the user side and decoder 1 on the base station side, and is used to feedback the base station-intelligent metasurface-user cascade channel H(0). Autoencoder 2 consists of encoder 2 on the user side and decoder 2 on the base station side, and is used to feedback the proportional vector p(t).

[0054] In an embodiment of the present invention, the encoder 1 is composed of a neural network and is used to compress the base station-intelligent metasurface-user channel H(0). The cascade channel H(0) is the input of the encoder 1. The output of the neural network of the encoder 1 is quantized to form a feedback bit stream s1(0), which is transmitted to the base station through the uplink control channel. The decoder 1 is composed of a neural network and is used to reconstruct the feedback cascade channel through the feedback bit stream s1(0). The input of the decoder 1 is the feedback bit stream s1(0), and the output is the reconstructed cascade channel.

[0055] In an embodiment of the present invention, the encoder 2 is composed of a neural network and is used to compress the proportional vector p(t). The proportional vector p(t) is the input of the encoder 2. The output of the neural network of the encoder 2 is quantized to form a feedback bit stream s2(t), which is transmitted to the base station via the uplink control channel. The decoder 2 is composed of a neural network and is used to reconstruct the feedback cascade channel using the feedback bit stream s2(t). The input of the decoder 2 is the feedback bit stream s2(t), and the output is the reconstructed proportional vector

[0056] The autoencoder 1 uses the input cascade channel H(t) and the reconstructed cascade channel The mean square error (MSE) between the two is used as the cost function for training to minimize the cost function. Since the gradient of the quantization operation is not differentiable, its gradient is set to a constant 1 during the training process; the cost function is described as follows:

[0057]

[0058] Autoencoder 2 uses the input scale vector p(t) and the reconstructed scale vector The MSE between is used as the cost function for training to minimize the cost function. Since the gradient of the quantization operation is not differentiable, its gradient is set to a constant 1 during the training process; the cost function is described as follows:

[0059]

[0060] The intelligent cascade channel feedback method for intelligent metasurface wireless communication of an embodiment of the present invention, when the base station-intelligent metasurface channel and the intelligent metasurface-user channel cannot be estimated separately, records the change ratio of the base station-each intelligent metasurface unit-user cascade channel compared with the previous moment in the form of a proportional vector, separates the intelligent metasurface-user channel information that needs frequent feedback from the cascade channel, and feeds back the base station-intelligent metasurface-user cascade channel and the proportional vector at different feedback intervals to reduce the channel feedback overhead and reduce the computational complexity by avoiding frequent processing of the base station-intelligent metasurface-user cascade channel.

[0061] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0062] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "N" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0063] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or N executable instructions for implementing a custom logical function or step of a process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

Claims

1. An intelligent cascade channel feedback method for intelligent metasurface wireless communication, characterized in that: The following steps are involved: Step 1: Construct two autoencoder neural networks, including autoencoder 1 and autoencoder 2, where autoencoder 1 consists of encoder 1 and decoder 1, and is used to feedback the base station-intelligent metasurface-user cascade channel. Autoencoder 2 consists of encoder 2 and decoder 2, and is used to feedback the ratio vector composed of the ratio of the base station-each intelligent metasurface unit-user cascade channel at adjacent moments. Step 2: During the coherence time of the base station-intelligent metasurface channel, when the channel feedback is performed for the first time, the user uses encoder 1 to compress the cascade channel H(0) into a feedback bit stream s1(0) and transmits it to the base station. After receiving the feedback bit stream s1(0), the base station uses decoder 1 to reconstruct the cascade channel. Step 3: During the second and subsequent channel feedback within the coherence time of the base station-intelligent metasurface channel, the user calculates the ratio between the base station-each intelligent metasurface unit-user cascade channel at the current moment and the base station-each intelligent metasurface unit-user cascade channel of the first feedback, forming a proportional vector p(t). The proportional vector p(t) is then compressed into a feedback bit stream s2(t) using encoder 2 and transmitted to the base station. After receiving the feedback bit stream S2(t), the base station reconstructs the proportional vector using decoder 2. And based on the cascade channel reconstructed from the first channel feedback With the proportional vector Calculate the reconstructed cascade channel at the current moment 2. The method according to claim 1, characterized in that The user is configured with a single antenna, and the cascade channel H(t) is composed of the product of the diagonal matrix consisting of the base station-smart metasurface channel matrix B(t) and the smart metasurface-user channel vector a(t), which can be described as follows: H(t)=diag(a(t))B(t) Where diag(·) is the transformation that converts a vector into a diagonal matrix.

3. The method according to claim 1, characterized in that The base station-i-th intelligent metasurface unit-user cascade channel is the i-th row h of the cascade channel matrix H(t) i (t), the i-th element p of the ratio vector p(t) i The calculation method of (t) is described as follows: Here, the superscript H denotes the conjugate transpose and ‖·‖2 is the Euclidean norm.

4. The method according to claim 3, characterized in that The cascade channel reconstructed from the first channel feedback With the proportional vector Calculate the reconstructed cascade channel at the current moment The method is described as follows: in, To reconstruct the scale vector The i-th element in .

5. The method according to claim 1, wherein Encoder 1 consists of a neural network and is used to compress the base station-intelligent metasurface-user cascade channel H(0). The cascade channel H(0) is the input of encoder 1. The output of the neural network of encoder 1 is quantized to form a feedback bit stream s1(0), which is transmitted to the base station through the uplink control channel.

6. The method according to claim 1, characterized in that Encoder 2 consists of a neural network and is used to compress the proportional vector p(t). The proportional vector p(t) is the input of encoder 2. The output of the neural network of encoder 2 is quantized to form a feedback bit stream s2(t), which is transmitted to the base station through the uplink control channel.

7. The method according to claim 1, characterized in that Decoder 1 consists of a neural network and is used to reconstruct the feedback cascade channel through the feedback bit stream s1(0). The input of decoder 1 is the feedback bit stream s1(0), and the output is the reconstructed cascade channel 8. The method according to claim 1, characterized in that Decoder 2 consists of a neural network and is used to reconstruct the feedback cascade channel through the feedback bit stream s2(t). The input of decoder 2 is the feedback bit stream s2(t), and the output is the reconstructed scale vector 9. The method according to claim 1, characterized in that The autoencoder 1 uses the input cascade channel H(t) and the reconstructed cascade channel The mean square error between is used as the cost function for training to minimize the cost function. During the training process, the gradient is set to a constant 1. The cost function is described as follows: where ‖·‖2 is the Euclidean norm.

10. The method according to claim 1, characterized in that Autoencoder 2 uses the input scale vector p(t) and the reconstructed scale vector The mean square error between is used as the cost function for training to minimize the cost function. During the training process, the gradient is set to a constant 1. The cost function is described as follows: where ‖·‖2 is the Euclidean norm.

Images