Communication method and device and storage medium

Abstract

The embodiment of the invention provides a communication method and device and a storage medium, relates to the technical field of communication, and is used for improving communication efficiency. The communication method comprises the following steps: acquiring a bit sequence, wherein the bit sequence comprises a plurality of bits; generating a complex symbol sequence based on the bit sequence and a preset mapping relation; wherein the complex symbol sequence comprises a plurality of complex symbols, and the preset mapping relation is used for indicating that the plurality of bits are mapped into the plurality of complex symbols; and sending the complex symbol sequence.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technologies, and in particular, to a communication method, apparatus, and storage medium. Background Art

[0002] In wireless communication, it is necessary to convert an information signal into a signal suitable for transmission over a channel for reliable transmission in a communication system. Currently, the modulation methods adopted in wireless communication all belong to two-dimensional modulation, in which multiple bits are mapped to one complex symbol during the modulation process, which can improve data transmission efficiency and spectral efficiency.

[0003] However, in two-dimensional modulation, since multiple bits are mapped to one complex symbol, more bandwidth resources are required to transmit information, reducing communication efficiency. Moreover, in two-dimensional modulation, information can only be transmitted in two dimensions, resulting in a relatively large number of symbols in each dimension and reducing spectral utilization, further limiting communication efficiency. Therefore, how to change the modulation method to improve communication efficiency is an urgent problem to be solved. Summary of the Invention

[0004] Embodiments of the present disclosure provide a communication method, apparatus, and storage medium for improving communication efficiency.

[0005] In a first aspect, a communication method is provided, including:

[0006] Obtaining a bit sequence, where the bit sequence includes multiple bits; generating a complex symbol sequence based on the bit sequence and a preset mapping relationship; where the complex symbol sequence includes multiple complex symbols, and the preset mapping relationship is used to indicate mapping multiple bits to multiple complex symbols; and sending the complex symbol sequence.

[0007] In a second aspect, a communication apparatus is provided, including:

[0008] A processing module, configured to obtain a bit sequence, where the bit sequence includes multiple bits.

[0009] The processing module is further configured to generate a complex symbol sequence based on the bit sequence and a preset mapping relationship; where the complex symbol sequence includes multiple complex symbols, and the preset mapping relationship is used to indicate mapping multiple bits to multiple complex symbols.

[0010] A communication module, configured to send the complex symbol sequence.

[0011] In a third aspect, a communication apparatus is provided, including a processor, and when the processor executes a computer program, the communication method of the first aspect is implemented.

[0012] Fourthly, a computer-readable storage medium is provided, which includes computer instructions; when the computer instructions are executed, the communication method in the first aspect is implemented.

[0013] In the embodiments of the present disclosure, a plurality of bits in a bit sequence are mapped to a plurality of complex symbols through a preset mapping relationship, rather than 1 complex symbol, thereby realizing high-dimensional modulation, reducing the bandwidth resources required for transmitting information, and improving communication efficiency. Moreover, information can be transmitted from multiple dimensions, reducing the number of symbols for transmitting information in each dimension, improving spectrum utilization, and further improving communication efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] To more clearly illustrate the technical solutions in the present disclosure, the following briefly introduces the drawings required to be used in some embodiments of the present disclosure. Obviously, the drawings in the following description are only the drawings of some embodiments of the present disclosure, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0015] Figure 1 Schematic diagram of the architecture of a communication system provided by an embodiment of the present disclosure;

[0016] Figure 2 Schematic diagram of the flowchart of a communication method provided by an embodiment of the present disclosure;

[0017] Figure 3 Schematic diagram of an end-to-end structure provided by an embodiment of the present disclosure;

[0018] Figure 4 Schematic diagram of the structure of a modulation neural network provided by an embodiment of the present disclosure;

[0019] Figure 5 Schematic diagram of the structure of a demodulation neural network provided by an embodiment of the present disclosure;

[0020] Figure 6 Flowchart of a modulation neural network and a demodulation neural network provided by an embodiment of the present disclosure;

[0021] Figure 7 Performance comparison diagram of high-dimensional modulation provided by an embodiment of the present disclosure;

[0022] Figure 8 Schematic diagram of the flowchart of another communication method provided by an embodiment of the present disclosure;

[0023] Figure 9 Schematic diagram of the structure of a communication device provided by an embodiment of the present disclosure;

[0024] Figure 10Schematic diagram of another communication device provided by an embodiment of the present disclosure. Detailed implementation manners

[0025] Next, the technical solutions in the embodiments of the present disclosure will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present disclosure.

[0026] In the description of the present disclosure, unless otherwise specified, " / " means "or". For example, A / B may represent A or B. The "and / or" herein is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, "at least one" means one or more, and "a plurality" means two or more. The words such as "first" and "second" do not limit the quantity and execution order, and the words such as "first" and "second" do not necessarily limit being different.

[0027] It should be noted that in the present disclosure, words such as "exemplary" or "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the present disclosure should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, using words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0028] As described in the background art, current modulation methods are all two-dimensional modulations. For example, Quadrature Amplitude Modulation (QAM) in the 5th Generation Mobile Communication Technology (5G), and Amplitude and Phase Shift Keying modulation (APSK) in the second-generation standard of Digital Video Broadcasting (DVB-S2). In two-dimensional modulation, multiple bits are mapped to 1 bit symbol, and information needs to be transmitted in two dimensions, resulting in a relatively large number of symbols in each dimension and low spectrum utilization, which limits the communication efficiency.

[0029] Based on this, the present disclosure provides a communication method. By mapping multiple bits in a bit sequence to multiple complex symbols through a preset mapping relationship instead of one complex symbol, high-dimensional modulation is achieved, reducing the bandwidth resources required for transmitting information and improving communication efficiency. Moreover, information can be transmitted from multiple dimensions, reducing the number of symbols for transmitting information in each dimension, improving the spectrum utilization rate, and further improving communication efficiency.

[0030] The communication method provided by the present disclosure can be applied to a communication system as Figure 1 shown. Figure 1 FIG. shows a schematic structural diagram of a communication system provided by an embodiment of the present disclosure. As Figure 1 shown, the communication system includes a first node 10 and a second node 20.

[0031] In a wireless communication scenario, the first node 10 communicates with the second node 20 through a wireless channel. For example, the first node 10 is a base station, the second node 20 is a terminal, and the base station communicates with the terminal through a wireless channel. For another example, the first node 10 is a terminal, the second node 20 is a wireless router, and the wireless router communicates with the terminal through a wireless channel. For another example, the first node 10 is a first base station, the second node 20 is a second base station, and the first base station communicates with the second base station through a wireless channel. For another example, the first node 10 is a first terminal, the second node 20 is a second terminal, and the first terminal communicates with the second terminal through a wireless channel. For another example, the first node 10 is a repeater, the second node 20 is a base station, and the base station communicates with the repeater through a wireless channel. For another example, the first node 10 is a terminal, the second node 20 is a repeater, and the repeater communicates with the terminal through a wireless channel. For another example, the first node 10 is a first repeater, the second node 20 is a second repeater, and the first repeater communicates with the second repeater through a wireless channel. For another example, the first node 10 is a base station, the second node 20 is a satellite, and the satellite communicates with the base station through a wireless channel. For another example, the first node 10 is a satellite, the second node 20 is a base station, and the base station communicates with the satellite through a wireless channel. For another example, the first node 10 is a terminal, the second node 20 is a satellite, and the satellite communicates with the terminal through a wireless channel. For another example, the first node 10 is a satellite, the second node 20 is a terminal, and the terminal communicates with the satellite through a wireless channel. For another example, the first node 10 is a ground device, the second node 20 is an aircraft, and the aircraft communicates with the ground device through a wireless channel. For another example, the first node 10 is a first aircraft, the second node 20 is a second aircraft, and the first aircraft communicates with the second aircraft through a wireless channel.

[0032] In the embodiment of the present disclosure, mainly taking the first node 10 as a base station and the second node 20 as a terminal as an example for illustration.

[0033] In some embodiments, the first node 10 is used to provide wireless access services for multiple terminals. Specifically, a base station provides a service coverage area (also referred to as a cell). Terminals entering this area can communicate with the base station via wireless signals to receive the wireless access services provided by the base station.

[0034] In some embodiments, the first node 10 may be a base station or an evolved base station (eNB or eNodeB) in Long Term Evolution (LTE), Long Term Evolution Advanced (LTE-A), a base station device in a 5G network, or a base station in a future communication system, etc. The base station may include various macro base stations, micro base stations, home base stations, remote radio heads, reconfigurable intelligent surfaces (RIS), routers, Wireless Fidelity (WIFI) devices and other network-side devices.

[0035] In some embodiments, the second node 20 may be a device with wireless transceiver functions, which can be deployed on land, including indoor or outdoor, handheld, wearable or vehicle-mounted; it can also be deployed on the water surface (such as a ship, etc.); it can also be deployed in the air (such as an airplane, a balloon, a satellite, etc.). The terminal may be a mobile phone, a tablet computer (Pad), a computer with wireless transceiver functions, a Virtual Reality (VR) terminal, an Augmented Reality (AR) terminal, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, etc. The embodiments of the present disclosure do not limit the application scenarios. The terminal is sometimes also referred to as a user, User Equipment (UE), access terminal, UE unit, UE station, mobile station, mobile unit, remote station, remote terminal, mobile device, UE terminal, wireless communication device, UE agent or UE device, etc., and the embodiments of the present disclosure do not limit this.

[0036] It should be noted that Figure 1 is only an exemplary framework diagram,Figure 1 The number of devices included, the names of the individual devices are not limited, and in addition to Figure 1 the devices shown, the communication system may also include other devices, such as core network devices.

[0037] The application scenarios of the embodiments of the present disclosure are not limited. The system architecture and service scenarios described in the embodiments of the present disclosure are for more clearly explaining the technical solutions of the embodiments of the present disclosure, and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. Those of ordinary skill in the art know that with the evolution of the network architecture and the emergence of new service scenarios, the technical solutions provided by the embodiments of the present disclosure are equally applicable to similar technical problems.

[0038] Figure 2 A flowchart showing a communication method provided by the present disclosure is shown, such as Figure 2 shown, the communication method is applied to a first node and includes the following steps:

[0039] S101. Obtain a bit sequence.

[0040] Wherein, the bit sequence includes a plurality of bits.

[0041] In some embodiments, the first node obtains the bit sequence based on the received sequence transmission information.

[0042] Wherein, the sequence transmission information includes the bit sequence to be transmitted and is used to instruct the first node to transmit the above bit sequence.

[0043] Exemplarily, the sequence transmission information may be sent to the first node by an upper-layer network device of the first node or a second node through the network. The bit sequence may be data generated by an upper-layer network device or a second node of the first node, or a control instruction sent by other network devices to the first node.

[0044] S102. Generate a complex symbol sequence based on the bit sequence and a preset mapping relationship.

[0045] Wherein, the complex symbol sequence includes a plurality of complex symbols, and the preset mapping relationship is used to indicate mapping a plurality of bits into a plurality of complex symbols.

[0046] It should be understood that a channel has certain frequency characteristics, and not all frequency components in a bit sequence can be transmitted through the channel. Therefore, it is necessary to modulate the bit sequence into a waveform suitable for transmission through the channel, and the sequence after modulation is called a complex symbol sequence. Moreover, modulation can convert the bit sequence into a more compact form, thereby reducing the bandwidth resources occupied during transmission. Additionally, modulation can also increase the anti-interference ability during transmission. In the embodiments of the present disclosure, a bit sequence is modulated into a complex symbol sequence through a preset mapping relationship to complete the subsequent transmission process.

[0047] Among them, the preset mapping relationship includes any one of the following: a modulation neural network, a modulation lookup table.

[0048] In some embodiments, by inputting a bit sequence into a modulation neural network, a complex symbol sequence corresponding to the bit sequence can be directly generated.

[0049] Exemplarily, by inputting the bit sequence b0, b1, b2,..., b M-1 into the modulation neural network, b0, b1, b2,..., b M-1 in which every Q m *N d / 2 bits can be mapped to N d / 2 complex symbols, and finally the complex symbol sequence S0, S1, S2,..., S (M-1) / Qm is output. Among them, Q m represents the modulation order, which is used to indicate the number of bits that each complex symbol can carry, and N d represents the modulation dimension, which is used to indicate the dimension when each complex symbol carries bits.

[0050] As a possible implementation manner, the modulation neural network is a neural network trained based on an end-to-end structure.

[0051] It should be noted that the end-to-end structure refers to a training method in the field of machine learning or deep learning that directly goes from the original input data to the final output result. In this training method, the entire neural network is trained to complete a specific task, rather than decomposing the task into multiple stages for separate training. This training method can make the neural network more automated and simplified, and can better adapt to different data and tasks, improving the performance and generalization ability of the neural network.

[0052] Among them, such as Figure 3As shown in the figure, it is a schematic diagram of an end-to-end structure provided by an embodiment of the present disclosure. The end-to-end structure includes at least one of the following: a modulation neural network and a demodulation neural network. In the end-to-end structure, bits are used as the input of the entire end-to-end structure and at the same time as the labels of the modulation neural network and the demodulation neural network. They are input into the modulation neural network, and the output result of the modulation neural network is input into the demodulation neural network through a channel. Finally, the demodulation neural network outputs the log-likelihood ratio (LLR). The binary cross-entropy (BCE) is used as the loss function for the entire process.

[0053] As Figure 4 shown, the modulation neural network includes at least one of the following: a first input layer, a segmentation layer, a symmetry constraint layer, a first network block, a coordinate transformation layer, a multiplication layer, and a power normalization layer.

[0054] In some embodiments, the first input layer is used to input a bit sequence into the segmentation layer.

[0055] In some embodiments, the segmentation layer is used to divide the bit sequence into a first bit sequence and a second bit sequence, input the first bit sequence into the symmetry constraint layer, and input the second bit sequence into the first network block.

[0056] Exemplarily, the first bit sequence is used to implement symmetry constraint, and the second bit sequence is used to calculate the amplitude. For example, the bit sequence input by the first input layer is 8 bits b0b1...b7. The segmentation layer divides the above bit sequence into a first bit sequence b0b1b2b3 and a second bit sequence b4b5b6b7. The first bit sequence b0b1b2b3 is input into the symmetry constraint layer, and the second bit sequence b4b5b6b7 is input into the first network block.

[0057] In some embodiments, the symmetry constraint layer is used to perform symmetry processing on the first bit sequence to generate a symmetry array and input the symmetry array into the multiplication layer.

[0058] Exemplarily, the symmetry constraint layer is used to map bit 0 in the first bit sequence to bit 1 and map bit 1 in the first bit sequence to bit -1. For example, when the first bit sequence is 0001, the symmetry array output by the symmetry constraint layer is [1, 1, 1, -1].

[0059] It should be noted that the symmetry constraint layer can be implemented by a fully connected layer, including a weight matrix and a bias vector. For example, when the input vector of the symmetry constraint layer is X = [x0, x1,...x Nd-1 , its output vector is Y = X * W + b. Where W = -2 * eye(N d-1), where W is the weight matrix and N d is a -1-dimensional unit diagonal matrix. The weight matrix represents the matrix composed of all weight parameters between two layers of neurons in the symmetric constraint layer. The weight matrix determines the connection strength between neurons, thereby affecting the learning and prediction capabilities of the modulation neural network. b = [1, 1,... 1], where b is the bias vector, and the bias vector is used to adjust the parameters of the activation function output of each layer of neurons. To ensure that the symmetric constraint layer remains unchanged during network training, the weight matrix and the bias vector need to remain unchanged during training.

[0060] In some embodiments, the first network block is used to generate a first array based on the second bit sequence and input the first array into the coordinate transformation layer.

[0061] Exemplarily, the first network block can be composed of multiple fully connected layers, can also be composed of a convolutional neural network, or can also be composed of a recurrent neural network.

[0062] In some embodiments, the coordinate transformation layer is used to convert the polar coordinate system corresponding to the first array into a rectangular coordinate system, generate a second array, and input the second array into the multiplication layer.

[0063] It should be noted that the diversity of the first array generated by the first network block is relatively low, and there may be a risk of overfitting. Therefore, coordinate transformation is required to perform a spatial transformation on the first data to improve the diversity of the first array and avoid the risk of overfitting, and finally generate a second array.

[0064] Exemplarily, the coordinate transformation layer can transform the coordinate representation [x0, x1,..., x Nd-1 in the first coordinate system corresponding to the first array to the coordinate representation [y0, y1,..., y Nd-1 in the second coordinate system. Taking the coordinate transformation layer converting a 4-dimensional polar coordinate system into a 4-dimensional rectangular coordinate system as an example, the specific coordinate transformation formula is as follows:

[0065] x1 = rcosθ1cosθ2cosθ3

[0066] x2 = rcosθ1cosθ2sinθ3

[0067] x3 = rcosθ1sinθ2

[0068] x4 = rsinθ1

[0069] In some embodiments, the multiplication layer is used to generate an initial complex number array based on the symmetric array and the second array and input the initial complex number array into the power normalization layer.

[0070] Exemplarily, with the modulation order Q m = 4 and the modulation dimension Nd Taking M = 4 as an example. Suppose the first bit sequence is 0000, then the symmetric array output by the symmetric constraint layer is [1, 1, 1, 1]. Suppose the second array output after the first network block and the coordinate transformation layer are calculated is [w, x, y, z], then the initial complex array generated by the multiplication layer based on the symmetric array and the second array is [w, x, y, z]. Another exemplary, suppose the first bit sequence is 0001, then the symmetric array output by the symmetric constraint layer is [1, 1, 1, -1]. Suppose the second array output after the first network block and the coordinate transformation layer are calculated is [w, x, y, z], then the initial complex array generated by the multiplication layer based on the symmetric array and the second array is [w, x, y, -z].

[0071] In some embodiments, the power normalization layer is used to normalize the initial complex array to generate a complex array and output the complex array.

[0072] Exemplary, the normalization process is used to normalize the mean of the initial complex array to 0 and the variance to 1, so that the average value of the output power is 1. The normalization process includes scaling the initial complex array proportionally and does not perform a translation process on the initial complex array. For example, taking the two initial complex arrays input in the above example as an example, [w, x, y, z] and [w, x, y, -z] are symmetric in the last dimension, then the complex array output by the power normalization layer is also symmetric in the last dimension, and the symmetry in other dimensions is similar to that in the last dimension, which will not be elaborated here.

[0073] Among them, the complex array is used to indicate the complex symbol sequence corresponding to the bit sequence.

[0074] Exemplary, taking the modulation order Q m = 4 and the modulation dimension N d = 4 as an example, when the complex array output by the power normalization layer is [w, x, y, z], the complex symbols indicated by the complex array are [S1, S2]. Among them, S1 = w + j * x, S2 = y + j * z, and j represents the unit imaginary number.

[0075] As Figure 5 shown, the demodulation neural network includes at least one of the following: a second input layer, a second network block, a coordinate inverse transformation layer, a third network block, and a splicing layer.

[0076] In some embodiments, the second input layer is used to input the complex array into the second network block and the coordinate inverse transformation layer.

[0077] In some embodiments, the second network block is used to calculate the first log-likelihood ratio for implementing symmetric constraints in the complex array and input the first log-likelihood ratio into the splicing layer.

[0078] Exemplarily, the second network block may be composed of multiple fully connected layers, or may be composed of a convolutional neural network, or may be composed of a recurrent neural network.

[0079] In some embodiments, the coordinate inverse transformation layer is used to convert the rectangular coordinate system corresponding to the complex number array into a polar coordinate system to obtain a third array, and input the third array into the third network block.

[0080] Exemplarily, the coordinate inverse transformation layer is used to transform the coordinate representation [y0, y1,..., y Nd-1 in the second coordinate system corresponding to the complex number array into the coordinate representation [x0, x1,..., x Nd-1 . Taking the coordinate transformation layer converting a 4D rectangular coordinate system into a 4D polar coordinate system as an example, the specific coordinate transformation formula is as follows:

[0081]

[0082]

[0083]

[0084]

[0085] In some embodiments, the third network block is used to calculate a second log-likelihood ratio based on the third array, and input the second log-likelihood ratio into the splicing layer.

[0086] Among them, the second log-likelihood ratio is used to characterize the log-likelihood ratio for calculating the amplitude in the complex number array.

[0087] Exemplarily, the third network block may be composed of multiple connection layers, or may be composed of a convolutional neural network, or may be composed of a recurrent neural network.

[0088] In some embodiments, the splicing layer is used to perform splicing processing on the first log-likelihood ratio and the second log-likelihood ratio to generate a target log-likelihood ratio, and output the target log-likelihood ratio.

[0089] Among them, the target log-likelihood ratio is used to characterize the relative possibility between each complex symbol in the complex symbol sequence that the second node can receive.

[0090] Such as Figure 6As shown, it is the flow structure diagram of the modulation neural network and the demodulation neural network provided by the embodiments of the present disclosure. Among them, in order to characterize the working process of the modulation neural network and the demodulation neural network during actual operation, the segmentation layer, the symmetry constraint layer, and the first network block in the modulation neural network are specifically presented in the form of a fully connected layer or a shaping layer, and the second network block, the coordinate inverse transformation layer, and the third network block in the demodulation neural network are specifically presented in the form of a fully connected layer or a shaping layer.

[0091] Specifically, in the modulation neural network, the input layer is equivalent to the first input layer in the present disclosure, the shaping layer (B, N, M) is equivalent to the segmentation layer in the present disclosure, the fully connected layer (B, N, 2) is equivalent to the symmetry constraint layer in the present disclosure, and the fully connected layer (B, N, M - 2) to the fully connected layer (B, 2 * N) + Sigmoid is equivalent to the first network block in the present disclosure. The coordinate transformation layer, the multiplication layer, and the power normalization layer are respectively equivalent to the coordinate transformation layer, the multiplication layer, and the power normalization layer in the present disclosure. In addition, a shaping layer (B, N, 2) is added between the coordinate transformation layer and the multiplication layer to align the dimensions of the complex symbols.

[0092] In the demodulation neural network, the symbol input layer is equivalent to the second input layer in the present disclosure, the coordinate inverse transformation layer is equivalent to the coordinate inverse transformation layer in the present disclosure, the fully connected layer (B, 32) + ELU to the shaping layer (B, N, M - 2) is equivalent to the third network block in the present disclosure, the fully connected layer (B, N, 16) + ELU to the fully connected layer (B, N, 2) is equivalent to the second network block in the present disclosure, and the splicing layer is equivalent to the splicing layer in the present disclosure. In addition, an SNR input layer is added above the symbol input layer to input the signal-to-noise ratio corresponding to the log-likelihood ratio LLR. A shaping layer and an absolute value layer are added between the symbol input layer and the coordinate inverse transformation layer to align the dimensions of the log-likelihood ratio. A shaping layer, a replication layer, and a multiplication layer are added after the splicing layer. The shaping layer is used to align the dimensions of the log-likelihood ratio, the replication layer is used to align the dimensions of the signal-to-noise ratio, and the multiplication layer is used to perform a weighted operation on the signal-to-noise ratio input by the replication layer and the log-likelihood ratio input by the shaping layer.

[0093] Exemplarily, Figure 6 The numbers in the brackets represent the output size of this layer. B represents the batch size, N is half of the modulation dimension, M is the modulation order. The coordinate transformation layer and the coordinate inverse transformation layer use polar coordinate to rectangular coordinate transformation and inverse transformation, and the activation function in the fully connected layer is the exponential linear unit function (ELU) or the sigmoid function.

[0094] As Figure 7 shown, it is the performance comparison diagram of high-dimensional modulation provided by the embodiments of the present disclosure. Figure 7During the training process of the modulation neural network under the conditions of modulation order 6, modulation dimension 4, and additive white Gaussian noise (AWGN) of 15 dB, the performance test of the block error rate (BLER) of the high-dimensional modulation of the modulation neural network is combined with the LDPC (low-density parity-check coding), and compared with the BLER performance of the geometrically shaped constellation diagram. From Figure 7 it can be seen that when the BLER is 0.1, the signal-to-noise ratio (SNR) corresponding to geometric shaping is 15.17 dB, and the SNR corresponding to high-dimensional modulation is 15.04 dB. The SNR required for high-dimensional modulation is lower than that for geometric shaping.

[0095] In some embodiments, based on the bit sequence and the modulation lookup table, the complex symbol sequence corresponding to the bit sequence can be directly determined.

[0096] As shown in Table 1, the modulation lookup table includes a plurality of complex symbols and a plurality of bits corresponding to the plurality of complex symbols.

[0097] Table 1

[0098] bit complex symbol 00000000 <![CDATA[S 0,1 ,S 0,2 > 00000001 <![CDATA[S 1,1 ,S 1,2 > ... ... 11111111 <![CDATA[S 255,1 ,S 255,2 >

[0099] As a possible implementation, the modulation lookup table is determined according to the modulation neural network.

[0100] Exemplarily, all possible combinations of Q m *N d / 2 bits are input into the modulation neural network to obtain the complex symbols corresponding to the bits of the possible combinations. For example, in the case of modulation order Q m = 4 and modulation dimension N d = 4, all possible combinations of 8 bits are 00000000, 00000001,..., 11111111, a total of 256 kinds. The above 256 bit combinations are input into the modulation neural network to obtain 256 complex arrays [w0, x0, y0, z0], [w1, x1, y1, z1],..., [w 255 , x 255 , y 255 , z 255 corresponding to the 256 bit combinations, and the 256 complex arrays are mapped to the corresponding complex symbols. For example, the complex symbol corresponding to the complex array [w i , x i , y i , z i is [Si,1 ,S i,2 . Finally, according to the complex symbols corresponding to the bits included in the bit sequence, the above complex symbols are combined to obtain a complex symbol sequence.

[0101] In another example, the bit sequence includes 00000000 and 00000001. From the above modulation lookup table, it can be seen that the complex symbol corresponding to 00000000 is S 0,1 ,S 0,2 , and the complex symbol corresponding to 00000001 is S 1,1 ,S 1,2 , then the complex symbol sequence corresponding to this bit sequence is S 0,1 ,S 0,2 ,S 1,1 ,S 1,2 . When the modulation lookup table is determined, based on the bit sequence and the modulation lookup table, the complex symbol sequence corresponding to the bit sequence can be directly determined without the participation of the modulation neural network.

[0102] S103. Transmit the complex symbol sequence.

[0103] As a possible implementation, after generating the complex symbol sequence, the first node can directly transmit the complex symbol sequence to the second node.

[0104] As another possible implementation, after the first node generates the complex symbol sequence, in response to receiving the sequence transmission instruction sent by the second node, the first node transmits the complex symbol sequence to the second node.

[0105] In this way, by presetting the mapping relationship, multiple bits in the bit sequence are mapped to multiple complex symbols instead of 1 complex symbol, achieving high-dimensional modulation, reducing the bandwidth resources required for transmitting information, and improving communication efficiency. Moreover, information can be transmitted from multiple dimensions, reducing the number of symbols for transmitting information in each dimension, improving the spectrum utilization rate, and further improving communication efficiency.

[0106] Figure 8 shows a schematic flowchart of a communication method provided by the present disclosure. As Figure 8 shown, this communication method is applied to the second node and includes the following steps:

[0107] S201. Receive the complex symbol sequence sent by the first node.

[0108] Among them, the complex symbol sequence includes multiple complex symbols.

[0109] S202. Determine the bit sequence based on the complex symbol sequence and the demodulation mapping relationship.

[0110] It should be understood that when the first node transmits a bit sequence, the bit sequence is modulated into a complex symbol sequence through a preset mapping relationship. Therefore, when the second node receives the complex symbol sequence, it is necessary to demodulate the complex symbol sequence into the corresponding bit sequence through the demodulation mapping relationship to obtain the corresponding data information.

[0111] Among them, the demodulation mapping relationship is used to indicate mapping multiple complex symbols into multiple bits. The demodulation mapping relationship includes any one of the following: a demodulation neural network, a demodulation lookup table.

[0112] As a possible implementation, the demodulation neural network is a neural network trained based on an end-to-end structure.

[0113] Among them, the end-to-end structure includes at least one of the following: a modulation neural network, a demodulation neural network.

[0114] As Figure 4 shown, the modulation neural network includes at least one of the following: a first input layer, a segmentation layer, a symmetry constraint layer, a first network block, a coordinate transformation layer, a multiplication layer, and a power normalization layer.

[0115] In some embodiments, the first input layer is used to input the bit sequence into the segmentation layer.

[0116] In some embodiments, the segmentation layer is used to divide the bit sequence into a first bit sequence and a second bit sequence, input the first bit sequence into the symmetry constraint layer, and input the second bit sequence into the first network block.

[0117] In some embodiments, the symmetry constraint layer is used to perform symmetry processing on the first bit sequence to generate a symmetry array, and input the symmetry array into the multiplication layer.

[0118] Among them, the symmetry constraint layer includes a weight matrix and a bias vector.

[0119] In some embodiments, the first network block is used to generate a first array based on the second bit sequence, and input the first array into the coordinate transformation layer.

[0120] In some embodiments, the coordinate transformation layer is used to convert the polar coordinate system corresponding to the first array into a rectangular coordinate system to generate a second array, and input the second array into the multiplication layer.

[0121] In some embodiments, the multiplication layer is used to generate an initial complex array based on the symmetry array and the second array, and input the initial complex array into the power normalization layer.

[0122] In some embodiments, the power normalization layer is used to perform normalization processing on the initial complex array to generate a complex array, and output the complex array.

[0123] AsFigure 5 As shown, the demodulation neural network includes at least one of the following: a second input layer, a second network block, a coordinate inverse transformation layer, a third network block, and a splicing layer.

[0124] In some embodiments, the second input layer is used to input a complex number array to the second network block and the coordinate inverse transformation layer.

[0125] In some embodiments, the second network block is used to calculate a first log-likelihood ratio for implementing symmetry constraints in the complex number array, and input the first log-likelihood ratio to the splicing layer.

[0126] In some embodiments, the coordinate inverse transformation layer is used to convert the rectangular coordinate system corresponding to the complex number array into a polar coordinate system to obtain a third array, and input the third array to the third network block.

[0127] In some embodiments, the third network block is used to calculate a second log-likelihood ratio based on the third array, and input the second log-likelihood ratio to the splicing layer.

[0128] In some embodiments, the splicing layer is used to perform a splicing process on the first log-likelihood ratio and the second log-likelihood ratio to generate a target log-likelihood ratio, and output the target log-likelihood ratio.

[0129] In some embodiments, the demodulation look-up table includes a plurality of bit sequences, and a plurality of complex symbols corresponding to each bit sequence in the plurality of bit sequences.

[0130] As a possible implementation, the demodulation look-up table is determined according to the demodulation neural network.

[0131] It should be noted that both the demodulation look-up table and the modulation look-up table include a plurality of bit sequences, and a plurality of complex symbols corresponding to each bit sequence in the plurality of bit sequences, and there is a corresponding relationship between the demodulation look-up table and the modulation look-up table. The relevant description of the demodulation look-up table can refer to the specific description of the modulation look-up table, and the present disclosure will not elaborate herein.

[0132] In this way, by presetting a mapping relationship to map a plurality of bits in the bit sequence to a plurality of complex symbols instead of 1 complex symbol, high-dimensional modulation is achieved, the bandwidth resources required for transmitting information are reduced, and the communication efficiency is improved. Moreover, information can be transmitted from multiple dimensions, the number of symbols for transmitting information in each dimension is reduced, the spectrum utilization rate is increased, and the communication efficiency is further improved.

[0133] It can be understood that, in order to implement the above functions, the communication device includes the corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should easily realize that, in combination with the algorithm steps of the examples described in the embodiments of the present disclosure, the present disclosure can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present disclosure.

[0134] The embodiments of the present disclosure can divide the functional modules of the communication device according to the above method embodiments. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one functional module. The above integrated module can be implemented in the form of hardware or software. It should be noted that the division of modules in the embodiments of the present disclosure is illustrative, only a logical function division, and there can be other division methods in actual implementation. The following takes the example of dividing each functional module corresponding to each function for illustration.

[0135] Figure 9 is a schematic structural diagram of a communication device applied to a first node provided by an embodiment of the present disclosure. The communication device 90 can execute the communication method provided by the above method embodiment. As Figure 9 shown, the communication device 90 includes a processing module 901 and a communication module 902.

[0136] The processing module 901 is configured to obtain a bit sequence, and the bit sequence includes a plurality of bits.

[0137] The processing module 901 is further configured to generate a complex symbol sequence based on the bit sequence and a preset mapping relationship; wherein, the complex symbol sequence includes a plurality of complex symbols, and the preset mapping relationship is used to indicate mapping a plurality of bits to a plurality of complex symbols.

[0138] The communication module 902 is configured to send the complex symbol sequence.

[0139] In some embodiments, the preset mapping relationship includes any one of the following: modulation neural network, modulation lookup table.

[0140] In some embodiments, the modulation neural network is a neural network trained based on an end-to-end structure.

[0141] In some embodiments, the end-to-end structure includes at least one of the following: modulation neural network, demodulation neural network.

[0142] In some embodiments, the modulation neural network includes at least one of the following: a first input layer, a segmentation layer, a symmetry constraint layer, a first network block, a coordinate transformation layer, a multiplication layer, and a power normalization layer; wherein, the first input layer is configured to input a bit sequence to the segmentation layer; the segmentation layer is configured to divide the bit sequence into a first bit sequence and a second bit sequence, input the first bit sequence to the symmetry constraint layer, and input the second bit sequence to the first network block, the first bit sequence is used to implement symmetry constraint, and the second bit sequence is used to calculate the amplitude; the symmetry constraint layer is configured to perform symmetry processing on the first bit sequence to generate a symmetry array, and input the symmetry array to the multiplication layer; the first network block is configured to generate a first array based on the second bit sequence, and input the first array to the coordinate transformation layer; the coordinate transformation layer is configured to convert the polar coordinate system corresponding to the first array into a rectangular coordinate system to generate a second array, and input the second array to the multiplication layer; the multiplication layer is configured to generate an initial complex number array based on the symmetry array and the second array, and input the initial complex number array to the power normalization layer; the power normalization layer is configured to perform normalization processing on the initial complex number array to generate a complex number array, and output the complex number array, and the complex number array is used to indicate the complex symbol sequence corresponding to the bit sequence.

[0143] In some embodiments, the symmetry constraint layer includes a weight matrix and a bias vector.

[0144] In some embodiments, the demodulation neural network includes at least one of the following: a second input layer, a second network block, a coordinate inverse transformation layer, a third network block, and a splicing layer; wherein, the second input layer is configured to input a complex number array to the second network block and the coordinate inverse transformation layer; the second network block is configured to calculate a first log-likelihood ratio for implementing symmetry constraint in the complex number array, and input the first log-likelihood ratio to the splicing layer; the coordinate inverse transformation layer is configured to convert the rectangular coordinate system corresponding to the complex number array into a polar coordinate system to obtain a third array, and input the third array to the third network block; the third network block is configured to calculate a second log-likelihood ratio based on the third array, and input the second log-likelihood ratio to the splicing layer, and the second log-likelihood ratio is used to represent the log-likelihood ratio for calculating the amplitude in the complex number array; the splicing layer is configured to perform splicing processing on the first log-likelihood ratio and the second log-likelihood ratio to generate a target log-likelihood ratio, and output the target log-likelihood ratio.

[0145] In some embodiments, the modulation look-up table includes a plurality of complex symbols, and a plurality of bits corresponding to each complex symbol in the plurality of complex symbols.

[0146] In some embodiments, the modulation look-up table is determined according to the modulation neural network.

[0147] In the case of implementing the functions of the above integrated modules in the form of hardware, embodiments of the present disclosure provide another possible structure of the communication device involved in the above embodiments. As Figure 10As shown, the communication device 100 includes: a processor 1002 and a bus 1004. Optionally, the communication device may further include a memory 1001; optionally, the communication device 100 may further include a communication interface 1003.

[0148] The processor 1002 may be a device that implements or executes various exemplary logical blocks, modules, and circuits described in connection with the embodiments of the present disclosure. The processor 1002 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute various exemplary logical blocks, modules, and circuits described in connection with the embodiments of the present disclosure. The processor 1002 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0149] The communication interface 1003 is used to connect to other devices through a communication network. The communication network may be an Ethernet, a radio access network, a wireless local area network (WLAN), etc.

[0150] The memory 1001 may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM), or other type of dynamic storage device that can store information and instructions. It may also be an electrically erasable programmable read-only memory (EEPROM), a magnetic disk storage medium, or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0151] As a possible implementation, the memory 1001 may exist independently of the processor 1002. The memory 1001 may be connected to the processor 1002 through the bus 1004 for storing instructions or program code. When the processor 1002 calls and executes the instructions or program code stored in the memory 1001, the communication method provided by the embodiments of the present disclosure can be implemented.

[0152] In another possible implementation, the memory 1001 may also be integrated with the processor 1002.

[0153] The bus 1004 can be an extended industry standard architecture (EISA) bus or the like. The bus 1004 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 10 only a thick line is used to represent it in Figure 10 , but it does not mean that there is only one bus or one type of bus.

[0154] Some embodiments of the present disclosure provide a computer-readable storage medium (e.g., a non-transitory computer-readable storage medium), in which computer program instructions are stored. When the computer program instructions run on a computer, the computer is caused to execute the communication method in any one of the above embodiments.

[0155] Exemplarily, the above computer-readable storage medium may include, but is not limited to: magnetic storage devices (such as hard disks, floppy disks, or magnetic tapes, etc.), optical discs (such as Compact Disks (CDs), Digital Versatile Disks (DVDs), etc.), smart cards, and flash memory devices (such as Erasable Programmable Read-Only Memories (EPROMs), cards, sticks, or key drives, etc.). The various computer-readable storage media described in the present disclosure may represent one or more devices and / or other machine-readable storage media for storing information. The term "machine-readable storage medium" may include, but is not limited to, wireless channels and various other media capable of storing, containing, and / or carrying instructions and / or data.

[0156] Embodiments of the present disclosure provide a computer program product containing instructions. When the computer program product runs on a computer, the computer is caused to execute the communication method described in any one of the above embodiments.

[0157] As described above, the above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present disclosure should be covered by the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A communication method, characterized in that, The method includes: Obtaining a bit sequence, where the bit sequence includes a plurality of bits; Generating a complex symbol sequence based on the bit sequence and a preset mapping relationship; wherein, the complex symbol sequence includes a plurality of complex symbols, and the preset mapping relationship is used to indicate mapping the plurality of bits to the plurality of complex symbols; Transmitting the complex symbol sequence.

2. The method according to claim 1, characterized in that, The preset mapping relationship includes any one of the following: a modulation neural network, a modulation look-up table.

3. The method according to claim 2, characterized in that, The modulation neural network is a neural network trained based on an end-to-end structure.

4. The method according to claim 3, wherein The end-to-end structure includes at least one of the following: the modulation neural network, the demodulation neural network.

5. The method according to claim 2, characterized in that, The modulation neural network includes at least one of the following: a first input layer, a segmentation layer, a symmetry constraint layer, a first network block, a coordinate transformation layer, a multiplication layer, and a power normalization layer; Wherein, the first input layer is used to input the bit sequence into the segmentation layer; The segmentation layer is used to divide the bit sequence into a first bit sequence and a second bit sequence, input the first bit sequence into the symmetry constraint layer, and input the second bit sequence into the first network block. The first bit sequence is used to implement symmetry constraint, and the second bit sequence is used to calculate the amplitude; The symmetry constraint layer is used to perform symmetry processing on the first bit sequence to generate a symmetry array, and input the symmetry array into the multiplication layer; The first network block is used to generate a first array based on the second bit sequence, and input the first array into the coordinate transformation layer; The coordinate transformation layer is used to convert the polar coordinate system corresponding to the first array into a rectangular coordinate system to generate a second array, and input the second array into the multiplication layer; The multiplication layer is used to generate an initial complex array based on the symmetry array and the second array, and input the initial complex array into the power normalization layer; The power normalization layer is used to perform normalization processing on the initial complex array to generate a complex array, and output the complex array. The complex array is used to indicate the complex symbol sequence corresponding to the bit sequence.

6. The method according to claim 5, wherein The symmetry constraint layer includes a weight matrix and a bias vector.

7. The method according to claim 4, wherein The demodulation neural network includes at least one of the following: a second input layer, a second network block, a coordinate inverse transformation layer, a third network block, and a splicing layer; Wherein, the second input layer is used to input the complex array into the second network block and the coordinate inverse transformation layer; The second network block is used to calculate a first log-likelihood ratio for implementing symmetry constraint in the complex array, and input the first log-likelihood ratio into the splicing layer; The coordinate inverse transformation layer is used to convert the rectangular coordinate system corresponding to the complex array into a polar coordinate system to obtain a third array, and input the third array into the third network block; The third network block is used to calculate a second log-likelihood ratio based on the third array, and input the second log-likelihood ratio into the splicing layer. The second log-likelihood ratio is used to characterize the log-likelihood ratio for calculating the amplitude in the complex array; The splicing layer is used to splice the first log-likelihood ratio and the second log-likelihood ratio to generate a target log-likelihood ratio and output the target log-likelihood ratio.

8. The method according to claim 2, characterized in that, The modulation look-up table includes a plurality of complex symbols and a plurality of bits corresponding to the plurality of complex symbols.

9. The method according to claim 2, characterized in that, The modulation look-up table is determined according to the modulation neural network.

10. A communication device, characterized in that, It includes a processor, and when the processor executes a computer program, it implements the communication method according to any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes computer instructions; wherein, when the computer instructions are executed, the communication method according to any one of claims 1 to 9 is implemented.

Images