Transmitting device, receiving device, system and method of terahertz wireless communication system
By using a sliding window bidirectional LSTM neural network to process terahertz signals, the problem of nonlinear distortion in terahertz communication is solved, achieving a balance between improving signal quality and computational efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2025-11-25
- Publication Date
- 2026-05-01
AI Technical Summary
In traditional terahertz communication systems, nonlinear distortion severely affects signal quality, and existing equalization schemes struggle to balance performance and computational complexity.
A sliding window bidirectional LSTM neural network is used for nonlinear equalization processing. The terahertz signal is processed by combining the initial random bit sequence generation with the variable step size sliding window mechanism of the bidirectional LSTM neural network.
It effectively alleviates nonlinear distortion in terahertz communication, reduces the error vector amplitude between the received signal and the ideal reference signal, and improves signal recovery.
Smart Images
Figure CN121966588A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of terahertz communication technology and related technical fields, specifically to a transmitting device, receiving device, system, and method for a terahertz wireless communication system. Background Technology
[0002] In recent years, with the increasing scarcity of microwave spectrum resources, the communications field has begun to expand towards higher frequency bands and wider spectrums. With the rising trend of carrier frequencies, terahertz (THz) technology, due to its unique advantages, is becoming a key enabling technology for next-generation wireless communication systems. The THz band not only provides ultra-wide bandwidth, supporting extremely fast data transmission speeds of hundreds of Gbps or even Tbps, but also achieves ultra-high sensing accuracy. Currently, with the rapid development of 5G global commercialization, next-generation mobile communication technologies are facing the challenge of increasingly diversified user demands. To realize the vision of ubiquitous connectivity, communication systems must achieve greater breakthroughs in transmission rates, latency control, and functional diversity. However, as communication evolves towards the terahertz band, the nonlinear distortion of high-power amplifiers, mixers, and other devices, combined with broadband signal memory effects, severely degrades signal quality and restricts transmission performance. To compensate for the impact of nonlinear distortion on signal transmission accuracy, the industry has proposed various wireless communication system architectures adapted to the terahertz band based on two core nonlinear equalization technologies.
[0003] However, traditional equalization schemes cannot adequately cover the needs of terahertz communication, and the nonlinear equalizers they use have their own limitations, making it difficult to achieve a balance between performance and computational complexity.
[0004] To address this issue, a new terahertz wireless communication system architecture based on a novel nonlinear equalizer is urgently needed. Summary of the Invention
[0005] The embodiments described herein provide a transmitting device, receiving device, system, and method for a terahertz wireless communication system, addressing problems existing in the prior art.
[0006] In a first aspect, according to the present disclosure, a transmitting device for a terahertz wireless communication system is provided, comprising: a transmitting end signal processing module and a transmitting module; The transmitting end signal processing module is configured to generate initial signal modulation symbols based on an initial random bit sequence, and to generate a transmitting end digital intermediate frequency signal based on the initial signal modulation symbols. The transmitting module is configured to convert the digital intermediate frequency signal of the transmitting end into an analog intermediate frequency signal of the transmitting end, and generate a terahertz signal based on the local oscillator signal and the analog intermediate frequency signal of the transmitting end.
[0007] In some embodiments of this disclosure, the transmitter signal processing module includes a sequence generation unit, a symbol mapping unit, an upsampling and pulse shaping unit, and an upconversion unit; The sequence generation unit is configured to generate an initial random bit sequence; The symbol mapping unit is configured to map the initial random bit sequence to an initial signal modulation symbol; The upsampling and pulse shaping unit is configured to upsample and pulse shape the initial signal modulation symbols to obtain a single-carrier baseband signal; The upconversion unit is configured to upconvert the single-carrier baseband signal to obtain a digital intermediate frequency signal for the transmitting end.
[0008] In some embodiments of this disclosure, the transmitting module includes a digital-to-analog converter unit, a local oscillator signal generation unit, a first mixer unit, a first amplifier unit, and a transmitting unit; The digital-to-analog converter is configured to convert the digital intermediate frequency signal from the transmitting end into an analog intermediate frequency signal from the transmitting end. The first local oscillator signal generating unit is configured to generate a local oscillator signal of a first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal; The first mixing unit is configured to mix the analog intermediate frequency signal from the transmitting end and the harmonic signal, and upconvert the signal to a terahertz signal. The first amplification unit is configured to amplify the terahertz signal; The transmitting unit is configured to radiate the amplified terahertz signal into space.
[0009] Secondly, according to the present disclosure, a receiving device for a terahertz wireless communication system is provided, comprising: a receiving end signal processing module and a receiving module; The receiving module is configured to receive terahertz signals, generate a receiving-end analog intermediate frequency signal based on the terahertz signals and the local oscillator signal, and convert the receiving-end analog intermediate frequency signal into a receiving-end digital intermediate frequency signal. The receiving end signal processing module is configured to process the receiving end digital intermediate frequency signal to obtain a baseband signal, and to perform resampling and retiming processing, linear equalization processing, frequency offset estimation processing and carrier phase recovery processing on the baseband signal to obtain a carrier phase synchronization signal, and to process and demap the carrier phase synchronization digital signal based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence.
[0010] In some embodiments of this disclosure, the receiving module includes a receiving unit, a second amplification unit, a second mixing unit, a second local oscillator signal generation unit, and an analog-to-digital conversion unit; The receiving unit is configured to receive terahertz signals; The second amplification unit is configured to amplify the received terahertz signal; The second local oscillator signal generating unit is configured to generate a local oscillator signal of the first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal; The second mixing unit is configured to perform heterodyne coherent reception of the received terahertz signal based on the frequency multiplication signal, and down-convert it to the analog intermediate frequency signal at the receiving end; The analog-to-digital conversion unit is configured to convert the analog intermediate frequency signal at the receiving end into a digital intermediate frequency signal at the receiving end.
[0011] In some embodiments of this disclosure, the receiver signal processing module includes a down-conversion unit, a resampling and retiming unit, a linear equalization unit, a frequency offset estimation unit, a carrier phase recovery unit, a nonlinear equalization unit, and a demapping unit; The downconversion unit is configured to perform downconversion processing on the digital intermediate frequency signal received to obtain a digital baseband signal; The resampling and retiming unit is configured to resample and retiming the digital baseband signal to obtain a first digital signal with matching sampling rate and synchronized timing. The linear equalization unit is configured to perform linear filtering on the first digital signal to obtain the second digital signal; The frequency offset estimation unit is configured to estimate the spectral offset of the second digital signal and compensate the second digital signal based on the spectral offset. The carrier phase recovery unit is configured to estimate the carrier phase offset of the second digital signal and compensate the second digital signal based on the carrier phase offset. The nonlinear equalization unit is configured to perform nonlinear equalization processing on the second digital signal after spectrum offset compensation and carrier phase offset compensation based on a sliding window bidirectional LSTM neural network to generate nonlinearly compensated target signal modulation symbols. The demapping unit is configured to demapping the nonlinearly compensated target signal modulation symbols to obtain a target random bit sequence.
[0012] In some embodiments of this disclosure, the sliding window bidirectional LSTM neural network includes an input layer, a BiLSTM layer, a linear layer, and an output layer; The input layer is configured to determine the first n and last n received symbols of the target time step based on the received symbols of the target time step, and generate a three-dimensional real-valued tensor based on the received symbols of the target time step and the first n and last n received symbols of the target time step, wherein the three-dimensional real-valued tensor includes 2n+1 symbols and I features and Q features of each symbol; The BiLSTM layer is configured to perform forward and backward calculations on the I and Q features of the 2n+1 symbols included in the three-dimensional real-valued tensor, respectively, to obtain the forward hidden state vector and the backward hidden state vector of the 2n+1 symbols. The linear layer is configured to determine the hidden state vector of each symbol based on the forward and backward hidden state vectors of the same symbol, and input the hidden state vectors of different symbols to two neurons respectively to determine the two-dimensional vector representation of the received symbol at the target time step. The output layer is configured to determine the target received symbol for the target time step based on a two-dimensional vector representation of the received symbol for the target time step.
[0013] In some embodiments of this disclosure, the output window length of the sliding window bidirectional LSTM neural network is smaller during training than the output window length during application.
[0014] Thirdly, according to the present disclosure, a terahertz wireless communication system is provided, including a transmitting device as described in any of the first aspects and a receiving device as described in any of the second aspects.
[0015] Fourthly, according to the present disclosure, a signal processing method for a terahertz wireless communication system is provided, comprising: An initial signal modulation symbol is generated based on an initial random bit sequence, and a digital intermediate frequency signal for the transmitting end is generated based on the initial signal modulation symbol. The digital intermediate frequency signal of the transmitting end is converted into an analog intermediate frequency signal of the transmitting end, and a terahertz signal is generated based on the local oscillator signal and the analog intermediate frequency signal of the transmitting end. Based on the terahertz signal and the local oscillator signal, an analog intermediate frequency (IF) signal for the receiving end is generated, and the analog IF signal for the receiving end is converted into a digital IF signal for the receiving end. The digital intermediate frequency signal at the receiving end is processed to obtain a digital baseband signal. The digital baseband signal is then resampled, retiming, linearly equalized, frequency offset estimated, and carrier phase recovered to obtain a carrier phase synchronization digital signal. Finally, the carrier phase synchronization digital signal is processed and demapped based on a sliding window bidirectional LSTM neural network to obtain an initial random bit sequence.
[0016] The terahertz wireless communication system transmitting device, receiving device, system, and method provided in this disclosure first generate initial signal modulation symbols based on an initial random bit sequence, and then generate a transmitting digital intermediate frequency (IF) signal based on the initial signal modulation symbols; next, convert the transmitting digital IF signal into a transmitting analog IF signal, and generate a terahertz signal based on the local oscillator signal and the transmitting analog IF signal; generate a receiving analog IF signal based on the terahertz signal and the local oscillator signal, and convert the receiving analog IF signal into a receiving digital IF signal; finally, process the receiving digital IF signal to obtain a digital baseband signal, and perform resampling, retiming, linear equalization, frequency offset estimation, and carrier phase recovery processing on the digital baseband signal to obtain a carrier phase synchronization digital signal; and finally, process and demap the carrier phase synchronization digital signal based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence. The target baseband signal is processed and demapped using a sliding window bidirectional LSTM neural network to obtain the target signal modulation symbol. On the one hand, the bidirectional structure of the sliding window bidirectional LSTM neural network can process the symbol sequence from both forward and backward directions simultaneously, effectively capturing the time dependence of the symbol sequence in the past and future contexts, greatly alleviating the nonlinear distortion caused by terahertz devices in terahertz communication, and reducing the error vector amplitude between the received signal and the ideal reference signal. On the other hand, a variable step size sliding window mechanism is introduced into the sliding window bidirectional LSTM neural network. Compared with the traditional inference method that only outputs the center symbol, this mechanism can complete the seamless reconstruction of the entire sequence with fewer output windows, which not only speeds up the inference speed but also enhances the training effect.
[0017] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description
[0018] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein: Figure 1 This is a schematic diagram of the structure of a transmitting device for a terahertz wireless communication system provided in an embodiment of this disclosure; Figure 2 This is a schematic diagram of the structure of a transmitting device for another terahertz wireless communication system provided in this embodiment of the present disclosure; Figure 3 This is a schematic diagram of the structure of a receiving device for a terahertz wireless communication system provided in an embodiment of this disclosure; Figure 4 This is a schematic diagram of the structure of a receiving device for another terahertz wireless communication system provided in this embodiment of the present disclosure; Figure 5 This is a schematic diagram of the structure of a sliding window bidirectional LSTM neural network provided in an embodiment of this disclosure; Figure 6 This is a schematic diagram of the structure of a terahertz wireless communication system provided in an embodiment of this disclosure; Figure 7 This is a schematic flowchart of a signal processing method for a terahertz wireless communication system provided in an embodiment of this disclosure.
[0019] In the accompanying diagram, markers with the same last two digits correspond to the same elements. It should be noted that the elements in the diagram are schematic and not drawn to scale. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.
[0021] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this subject matter pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the specification and in the relevant art, and shall not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, the statement of “connecting” or “coupling” two or more parts together shall mean that these parts are directly joined together or joined through one or more intermediate components.
[0022] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0023] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists, A and B exist simultaneously, or B exists. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0024] Furthermore, in all embodiments of this disclosure, terms such as “first” and “second” are used only to distinguish one component (or part of a component) from another component (or another part of a component).
[0025] In the description of this application, unless otherwise stated, "multiple" means two or more (including two), and similarly, "multiple groups" means two or more (including two groups).
[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0027] In view of the problems existing in the prior art, the present disclosure provides a transmitting device for a terahertz wireless communication system. Figure 1 This is a schematic diagram of the structure of a transmitting device for a terahertz wireless communication system provided in an embodiment of this disclosure, as shown below. Figure 1 As shown, the transmitting device of the terahertz wireless communication system includes: a transmitting end signal processing module 10 and a transmitting module 20; wherein, the transmitting end signal processing module 10 is configured to generate initial signal modulation symbols based on an initial random bit sequence, and generate a transmitting end digital intermediate frequency signal based on the initial signal modulation symbols; the transmitting module 20 is configured to convert the transmitting end digital intermediate frequency signal into a transmitting end analog intermediate frequency signal, and generate a terahertz signal based on the local oscillator signal and the transmitting end analog intermediate frequency signal.
[0028] The signal modulation symbols include QPSK symbols, 8QAM symbols, 16QAM symbols, 64QAM symbols, etc., and the embodiments disclosed herein do not specifically limit them.
[0029] Specifically, in combination Figure 2 The transmitter signal processing module 10 includes a sequence generation unit 11, a symbol mapping unit 12, an upsampling and pulse shaping unit 13, and an upconversion unit 14. The sequence generation unit 11 is configured to generate an initial random bit sequence. The symbol mapping unit 12 is configured to map the initial random bit sequence to an initial signal modulation symbol. The upsampling and pulse shaping unit 13 is configured to upsample and pulse shape the initial signal modulation symbol to obtain a single-carrier baseband signal. The upconversion unit 14 is configured to upconvert the single-carrier baseband signal to obtain a transmitter digital intermediate frequency signal.
[0030] The transmitting module 20 includes a digital-to-analog conversion unit 21, a local oscillator signal generation unit 22, a first mixing unit 23, a first amplification unit 24, and a transmitting unit 25. The digital-to-analog conversion unit 21 is configured to convert the digital intermediate frequency signal of the transmitting end into an analog intermediate frequency signal of the transmitting end. The first local oscillator signal generation unit 22 is configured to generate a local oscillator signal of a first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal. The first mixing unit 23 is configured to mix the analog intermediate frequency signal of the transmitting end and the frequency multiplier signal, and up-convert the frequency to a terahertz signal. The first amplification unit 24 is configured to amplify the terahertz signal. The transmitting unit 25 is configured to radiate the amplified terahertz signal into space.
[0031] Specifically, in the transmitter signal processing module, the sequence generation unit generates an initial random bit sequence based on the Mersenne Twister algorithm (MTRS), and then maps it to signal modulation symbols. The sequence generation unit uses the initial random bit sequence generated by the Mersenne Twister algorithm instead of a pseudo-random bit sequence, which can effectively eliminate data periodicity and avoid overfitting. The symbol mapping unit maps the initial random bit sequence to initial signal modulation symbols. After symbol mapping, the upsampling and pulse shaping unit upsamples the signal and uses a raised cosine (RC) filter for pulse shaping to obtain a single-carrier baseband signal. Finally, the upconversion unit upconverts the single-carrier baseband signal to a transmitter digital intermediate frequency signal and sends it to the digital-to-analog converter unit of the transmitter module.
[0032] In a specific example, in the transmitting module, a digital-to-analog converter (DAC) converts the digital intermediate frequency (IF) signal from the transmitting end into an analog IF signal. Simultaneously, a first local oscillator (LO) signal generation unit generates a 17.5 GHz sine wave. This sine wave is multiplied by a frequency multiplier of six to obtain a 105 GHz multiplied signal. This signal drives a first mixer unit (which includes a terahertz subharmonic mixer). The first mixer unit mixes the obtained analog IF signal with the 105 GHz multiplied signal from the first LO signal generation unit, up-converting it to a 220 GHz terahertz signal. The 220 GHz terahertz signal is first amplified by a first amplification unit (which includes a terahertz low-noise amplifier) and then radiated into space through a transmitting unit (which includes a terahertz horn antenna).
[0033] The transmitter of the terahertz wireless communication system adopts a two-stage signal upconversion method. First, the digital signal from baseband to intermediate frequency is upconverted in the signal processing module of the transmitter. Then, the analog signal from intermediate frequency to terahertz (THz) is upconverted in the first mixing unit of the transmitter module to generate the terahertz signal.
[0034] The terahertz wireless communication system transmitting device provided in this embodiment generates an initial signal modulation symbol based on an initial random bit sequence and generates a digital intermediate frequency (IF) signal based on the initial signal modulation symbol. The transmitting module converts the IF signal into an analog IF signal and generates a terahertz signal based on the local oscillator signal and the analog IF signal. The generated initial signal modulation symbol is obtained by mapping the initial random bit sequence generated by the Mason twitch algorithm. Compared with the prior art that generates the signal modulation sequence based on a pseudo-random bit sequence, this method can better eliminate the periodicity of the data and avoid overfitting.
[0035] Based on the above embodiments, this disclosure also provides a receiving device for a terahertz wireless communication system. Figure 3 This is a schematic diagram of the structure of a receiving device for a terahertz wireless communication system provided in an embodiment of this disclosure, as shown below. Figure 3 As shown, the receiving device of the terahertz wireless communication system includes: a receiving signal processing module 40 and a receiving module 30; the receiving module 30 is configured to receive terahertz signals, generate a receiving analog intermediate frequency signal based on the terahertz signal and the local oscillator signal, and convert the receiving analog intermediate frequency signal into a receiving digital intermediate frequency signal; the receiving signal processing module 40 is configured to process the receiving digital intermediate frequency signal to obtain a baseband signal, and perform resampling and retiming processing, linear equalization processing, frequency offset estimation processing and carrier phase recovery processing on the baseband signal to obtain a carrier phase synchronization signal, and process and demap the carrier phase synchronization signal based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence.
[0036] LSTM stands for Long Short-Term Memory, which is a network of long short-term memory.
[0037] The receiving module 30 includes a receiving unit 31, a second amplification unit 32, a second mixing unit 33, a second local oscillator signal generation unit 34, and an analog-to-digital conversion unit 35. The receiving unit 31 is configured to receive terahertz signals. The second amplification unit 32 is configured to amplify the received terahertz signals. The second local oscillator signal generation unit 34 is configured to generate a local oscillator signal of a first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal. The second mixing unit 33 is configured to perform heterodyne coherent reception of the received terahertz signals based on the frequency multiplier signal and down-convert the signals to receive analog intermediate frequency signals. The analog-to-digital conversion unit 35 is configured to convert the received analog intermediate frequency signals into received digital intermediate frequency signals.
[0038] The receiver signal processing module 40 includes a down-conversion unit 41, a resampling and retiming unit 42, a linear equalization unit 43, a frequency offset estimation unit 44, a carrier phase recovery unit 45, a nonlinear equalization unit 46, and a demapping unit 47. The down-conversion unit 41 is configured to down-convert the received digital intermediate frequency signal to obtain a digital baseband signal. The resampling and retiming unit 42 is configured to resample and re-timing the digital baseband signal to obtain a first digital signal with matching sampling rate and synchronized timing. The linear equalization unit 43 is configured to perform linear filtering on the first digital signal to obtain a second digital signal. The frequency offset estimation unit 44... 4 is configured to estimate the spectral offset of the second digital signal and compensate the second digital signal based on the spectral offset; carrier phase recovery unit 45 is configured to estimate the carrier phase offset of the second digital signal and compensate the second digital signal based on the carrier phase offset; nonlinear equalization unit 46 is configured to perform nonlinear equalization processing on the second digital signal after spectral offset compensation and carrier phase offset compensation based on a sliding window bidirectional LSTM neural network to generate nonlinearly compensated target signal modulation symbols; demapping unit 47 is configured to demapping the nonlinearly compensated target signal modulation symbols to obtain a target random bit sequence.
[0039] In a specific example, the terahertz signal transmitted by the transmitter of the terahertz wireless communication system is captured by the receiving unit of the receiving module of the receiving device of the terahertz wireless communication system after wireless transmission. The received terahertz signal is amplified by a second amplification unit (the second amplification unit includes a terahertz low-noise amplifier) and then sent to a second mixing unit. Simultaneously, a second local oscillator signal generation unit generates a 17.5 GHz sine wave. This sine wave is multiplied by a frequency multiplier of 6 to obtain a 105 GHz multiplied signal, which is then sent to the second mixing unit. The second mixing unit uses the 105 GHz multiplied signal to perform heterodyne coherent reception of the received terahertz signal, down-converting it to a 10 GHz analog intermediate frequency (IF) signal for the receiving end. The down-converted 10 GHz analog IF signal for the receiving end is then sent to an analog-to-digital converter (ADC) to obtain a digital IF signal for the receiving end.
[0040] The digital intermediate frequency (IF) signal at the receiving end undergoes down-conversion within the receiving end signal processing module to obtain a digital baseband signal. Next, the digital baseband signal is resampled and retiming processed to obtain a first digital signal with matched sampling rate and synchronized timing. Then, a linear equalization unit performs linear filtering on the first digital signal using the constant modulus algorithm (CMA) to obtain a second digital signal. A frequency offset estimation unit estimates the spectral offset of the second digital signal and compensates for it based on the spectral offset. A carrier phase recovery unit estimates the carrier phase offset of the second digital signal and compensates for it based on the carrier phase offset. Finally, a nonlinear equalization unit uses a sliding window bidirectional LSTM neural network to perform nonlinear equalization on the second digital signal after spectral offset and carrier phase offset compensation to generate a nonlinearly compensated target signal modulation symbol. Finally, a demapping unit demaps the nonlinearly compensated target signal modulation symbol to obtain the target random bit sequence, thus achieving signal recovery.
[0041] The receiving device of the terahertz wireless communication system adopts a two-stage signal downconversion method based on heterodyne detection. First, the terahertz signal is downconverted to a 10GHz analog intermediate frequency signal in the second mixing unit of the receiving signal processing module, completing the first stage of analog signal downconversion. The generated receiving digital intermediate frequency signal is sent to the receiving signal processing module. The receiving digital center signal is downconverted by the downconversion unit of the receiving signal processing module to obtain the digital baseband signal.
[0042] As a specific implementation, the receiving device of the terahertz wireless communication system provided in this embodiment of the present disclosure has a nonlinear equalization unit of the receiving end signal processing module that performs nonlinear equalization processing on the carrier phase synchronization digital signal based on a sliding window bidirectional LSTM neural network.
[0043] Among them, such as Figure 5As shown, the sliding window bidirectional LSTM neural network includes an input layer, a Bi-directional Long Short-Term Memory (BiLSTM) layer, a linear layer, and an output layer. The input layer is configured to determine the nth and nth received symbols at the target time step based on the received symbols at that time step, and generate a three-dimensional real-valued tensor based on the received symbols at the target time step and the nth and nth received symbols at the target time step. The three-dimensional real-valued tensor includes 2n+1 symbols and the I and Q features of each symbol. The BiLSTM layer... The STM layer is configured to sequentially perform forward and backward computations on the I and Q features of the 2n+1 symbols in the 3D real-valued tensor to obtain the forward and backward hidden state vectors of the 2n+1 symbols; the linear layer is configured to determine the hidden state vector of each symbol based on the forward and backward hidden state vectors of the same symbol, and input the hidden state vectors of different symbols into two neurons respectively to determine the two-dimensional vector representation of the received symbol at the target time step; the output layer is configured to determine the target received symbol at the target time step based on the two-dimensional vector representation of the received symbol.
[0044] The sliding window bidirectional LSTM neural network consists of an input layer, a BiLSTM layer, a linear layer, and an output layer. It comprises two key components: LSTM units and a bidirectional structure. The first layer is the input layer. For a received symbol Xt with time step t, an input window is constructed by concatenating the current symbol with its n preceding and n following symbols to learn the memory effect: X(t) = [xt-n,...,xt,...,xt+n]. The input window length is Lin = 2n+1. Each symbol contains two features: I-features and Q-features. The entire received sequence is fed as a three-dimensional real-valued tensor into the BiLSTM layer of the sliding window bidirectional LSTM neural network, with a shape of (B, Lin, ...). 2), where B is the batch size, Lin is the input window length, and 2 indicates that each symbol includes two features; the second layer is a BiLSTM layer. Since training LSTM is difficult, the number of recurrent layers is generally no more than 3. To reduce complexity, a single-layer BiLSTM network is used. The bidirectional process of the BiLSTM network enriches the feature representation. The BiLSTM layer sequentially performs forward and backward calculations on the I and Q features of the 2n+1 symbols included in the three-dimensional real-valued tensor to obtain the forward hidden state vector and backward hidden state vector of the 2n+1 symbols; the linear layer first determines the hidden state vector of each symbol based on the forward and backward hidden state vectors of the same symbol, and then inputs the hidden state vectors of different symbols into two neurons respectively. Based on the neurons, the two-dimensional vector representation of the received symbol at each time step is determined. That is, the linear layer first performs the hidden state vector calculation, and then completely connects the hidden state vector of each received symbol to the two neurons of the linear layer, mapping the high-dimensional hidden features of the received symbols at each time step to a two-dimensional vector. Finally, the output layer determines the target received symbol for each time step based on the two-dimensional vector representation of the received symbol at each time step, and finally determines the target signal modulation symbol based on the target received symbols at all target time steps.
[0045] This bidirectional LSTM neural network employs a variable step size sliding window mechanism, meaning that the output window length during training is smaller than the output window length during application, reducing the complexity of the inference process and improving the inference speed.
[0046] The receiving device for a terahertz wireless communication system provided in this disclosure includes a receiving module that receives a terahertz signal and generates a receiving-end analog intermediate frequency (IF) signal based on the terahertz signal and a local oscillator signal, and converts the receiving-end analog IF signal into a receiving-end digital IF signal. A receiving-end signal processing module processes the receiving-end digital IF signal to obtain a baseband signal, and performs resampling and retiming processing, linear equalization processing, frequency offset estimation processing, and carrier phase recovery processing on the baseband signal to obtain a carrier phase synchronization signal. Finally, a target random bit sequence is obtained by performing nonlinear equalization processing on the carrier phase synchronization signal based on a sliding window bidirectional LSTM neural network and demapping it. The innovative BiLSTM design in the nonlinear equalization process introduces a variable step-size sliding window mechanism, which can achieve seamless reconstruction of the entire sequence with fewer output windows, effectively improving inference speed. Furthermore, by using a sliding window bidirectional LSTM neural network, nonlinear distortion can be effectively suppressed during signal processing, significantly reducing EVM (Error Vector Magnitude), making the symbol points of the equalized signal constellation diagram more concentrated, and effectively recovering the original signal.
[0047] Based on the above embodiments, this disclosure also provides a terahertz wireless communication system, such as... Figure 6 As shown, the terahertz wireless communication system includes the transmitting device and receiving device described in any of the above embodiments, and has the beneficial effects described in any of the above embodiments. This disclosure will not provide a specific description of these effects.
[0048] Based on the above embodiments, this disclosure also provides a signal processing method for a terahertz wireless communication system, such as... Figure 7 As shown, the signal processing methods for terahertz wireless communication systems include: S110. Generate initial signal modulation symbols based on the initial random bit sequence, and generate the transmitter digital intermediate frequency signal based on the initial signal modulation symbols.
[0049] S120. Convert the digital intermediate frequency signal at the transmitting end into an analog intermediate frequency signal at the transmitting end, and generate a terahertz signal based on the local oscillator signal and the analog intermediate frequency signal at the transmitting end.
[0050] S130. Generate the analog intermediate frequency signal for the receiver based on the terahertz signal and the local oscillator signal, and convert the analog intermediate frequency signal for the receiver into a digital intermediate frequency signal for the receiver.
[0051] S140. The digital intermediate frequency signal at the receiving end is processed to obtain a digital baseband signal. The digital baseband signal is then resampled, retiming, linearly equalized, frequency offset estimated, and carrier phase recovered to obtain a carrier phase synchronization digital signal. The carrier phase synchronization digital signal is then processed and demapped based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence.
[0052] The signal processing method for a Hertz wireless communication system provided in this disclosure first generates an initial signal modulation symbol based on an initial random bit sequence, and then generates a transmitting digital intermediate frequency (IF) signal based on the initial signal modulation symbol. Next, the transmitting IF signal is converted into a transmitting analog IF signal, and a terahertz signal is generated based on the local oscillator signal and the transmitting analog IF signal. Based on the terahertz signal and the local oscillator signal, a receiving analog IF signal is generated, and the receiving analog IF signal is converted into a receiving digital IF signal. Finally, the receiving digital IF signal is processed to obtain a digital baseband signal, and the digital baseband signal is subjected to resampling, retiming, linear equalization, frequency offset estimation, and carrier phase recovery processing to obtain a carrier phase synchronization digital signal. Finally, the carrier phase synchronization digital signal is processed and demapped based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence. The target baseband signal is processed and mapped to obtain the target signal modulation symbol by a sliding window bidirectional LSTM neural network. On the one hand, the bidirectional structure of the sliding window bidirectional LSTM neural network can process the symbol sequence from both forward and backward directions at the same time, effectively capturing the time dependence of the symbol sequence in the past and future contexts, greatly alleviating the nonlinear distortion caused by terahertz devices in terahertz communication, and reducing the error vector amplitude between the received signal and the ideal reference signal. On the other hand, a variable step size sliding window mechanism is introduced into the sliding window bidirectional LSTM neural network. Compared with the traditional inference method that only outputs the center symbol, this mechanism can complete the seamless reconstruction of the entire sequence with fewer output windows, which not only speeds up the inference speed but also enhances the training effect.
[0053] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0054] In the various embodiments of this application, the functional units or modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0055] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0056] Unless otherwise expressly indicated by the context, the singular form of words used herein and in the appended claims includes the plural form, and vice versa. Thus, when referring to the singular, the plural form of the corresponding term is generally included. Similarly, the terms “comprising” and “including” shall be interpreted as including rather than exclusively. Likewise, the terms “including” and “or” shall be interpreted as including unless such interpretation is expressly prohibited herein. Where the term “example” is used herein, particularly when it follows a set of terms, the “example” is merely exemplary and illustrative and should not be considered exclusive or extensive.
[0057] Further aspects and scope of adaptation become apparent from the description provided herein. It should be understood that various aspects of this application may be implemented individually or in combination with one or more other aspects. It should also be understood that the descriptions and specific embodiments herein are for illustrative purposes only and are not intended to limit the scope of this application.
[0058] Several embodiments of this disclosure have been described in detail above. However, it is obvious that those skilled in the art can make various modifications and variations to the embodiments of this disclosure without departing from the spirit and scope of this disclosure. The scope of protection of this disclosure is defined by the appended claims.
Claims
1. A transmitting device for a terahertz wireless communication system, characterized in that, include: Transmitter signal processing module and transmitter module; The transmitting end signal processing module is configured to generate initial signal modulation symbols based on an initial random bit sequence, and to generate a transmitting end digital intermediate frequency signal based on the initial signal modulation symbols. The transmitting module is configured to convert the digital intermediate frequency signal of the transmitting end into an analog intermediate frequency signal of the transmitting end, and generate a terahertz signal based on the local oscillator signal and the analog intermediate frequency signal of the transmitting end.
2. The launching device according to claim 1, characterized in that, The transmitter signal processing module includes a sequence generation unit, a symbol mapping unit, an upsampling and pulse shaping unit, and an upconversion unit. The sequence generation unit is configured to generate an initial random bit sequence; The symbol mapping unit is configured to map the initial random bit sequence to an initial signal modulation symbol; The upsampling and pulse shaping unit is configured to upsample and pulse shape the initial signal modulation symbols to obtain a single-carrier baseband signal; The upconversion unit is configured to upconvert the single-carrier baseband signal to obtain a digital intermediate frequency signal for the transmitting end.
3. The launching device according to claim 1, characterized in that, The transmitting module includes a digital-to-analog conversion unit, a first local oscillator signal generation unit, a first mixing unit, a first amplification unit, and a transmitting unit; The digital-to-analog converter is configured to convert the digital intermediate frequency signal from the transmitting end into an analog intermediate frequency signal from the transmitting end. The first local oscillator signal generating unit is configured to generate a local oscillator signal of a first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal; The first mixing unit is configured to mix the analog intermediate frequency signal from the transmitting end and the harmonic signal, and upconvert the signal to a terahertz signal. The first amplification unit is configured to amplify the terahertz signal; The transmitting unit is configured to radiate the amplified terahertz signal into space.
4. A receiving device for a terahertz wireless communication system, characterized in that, include: Receiver signal processing module and receiver module; The receiving module is configured to receive terahertz signals, generate a receiving-end analog intermediate frequency signal based on the terahertz signals and the local oscillator signal, and convert the receiving-end analog intermediate frequency signal into a receiving-end digital intermediate frequency signal. The receiving end signal processing module is configured to process the receiving end digital intermediate frequency signal to obtain a baseband signal, and to perform resampling and retiming processing, linear equalization processing, frequency offset estimation processing and carrier phase recovery processing on the baseband signal to obtain a carrier phase synchronization signal, and to process and demap the carrier phase synchronization digital signal based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence.
5. The receiving device according to claim 4, characterized in that, The receiving module includes a receiving unit, a second amplification unit, a second mixing unit, a second local oscillator signal generation unit, and an analog-to-digital conversion unit; The receiving unit is configured to receive terahertz signals; The second amplification unit is configured to amplify the received terahertz signal; The second local oscillator signal generating unit is configured to generate a local oscillator signal of the first frequency and amplify the local oscillator signal by frequency multiplication to obtain a frequency multiplier signal; The second mixing unit is configured to perform heterodyne coherent reception of the received terahertz signal based on the frequency multiplication signal, and down-convert it to the analog intermediate frequency signal at the receiving end; The analog-to-digital conversion unit is configured to convert the analog intermediate frequency signal at the receiving end into a digital intermediate frequency signal at the receiving end.
6. The receiving device according to claim 4, characterized in that, The receiver signal processing module includes a down-conversion unit, a resampling and retiming unit, a linear equalization unit, a frequency offset estimation unit, a carrier phase recovery unit, a nonlinear equalization unit, and a demapping unit; The downconversion unit is configured to perform downconversion processing on the digital intermediate frequency signal received to obtain a digital baseband signal; The resampling and retiming unit is configured to resample and retiming the digital baseband signal to obtain a first digital signal with matching sampling rate and synchronized timing. The linear equalization unit is configured to perform linear filtering on the first digital signal to obtain the second digital signal; The frequency offset estimation unit is configured to estimate the spectral offset of the second digital signal and compensate the second digital signal based on the spectral offset. The carrier phase recovery unit is configured to estimate the carrier phase offset of the second digital signal and compensate the second digital signal based on the carrier phase offset. The nonlinear equalization unit is configured to perform nonlinear equalization processing on the second digital signal after spectrum offset compensation and carrier phase offset compensation based on a sliding window bidirectional LSTM neural network to generate nonlinearly compensated target signal modulation symbols. The demapping unit is configured to demapping the nonlinearly compensated target signal modulation symbols to obtain a target random bit sequence.
7. The receiving device according to claim 6, characterized in that, The sliding window bidirectional LSTM neural network includes an input layer, a BiLSTM layer, a linear layer, and an output layer. The input layer is configured to determine the first n and last n received symbols of the target time step based on the received symbols of the target time step, and generate a three-dimensional real-valued tensor based on the received symbols of the target time step and the first n and last n received symbols of the target time step, wherein the three-dimensional real-valued tensor includes 2n+1 symbols and I features and Q features of each symbol; The BiLSTM layer is configured to perform forward and backward calculations on the I and Q features of the 2n+1 symbols included in the three-dimensional real-valued tensor, respectively, to obtain the forward hidden state vector and the backward hidden state vector of the 2n+1 symbols. The linear layer is configured to determine the hidden state vector of each symbol based on the forward and backward hidden state vectors of the same symbol, and input the hidden state vectors of different symbols to two neurons respectively to determine the two-dimensional vector representation of the received symbol at the target time step. The output layer is configured to determine the target received symbol for the target time step based on a two-dimensional vector representation of the received symbol for the target time step.
8. The receiving device according to claim 7, characterized in that, The output window length of the sliding window bidirectional LSTM neural network is smaller during training than during application.
9. A terahertz wireless communication system, characterized in that, It includes the transmitting device as described in any one of claims 1-3 and the receiving device as described in any one of claims 4-8.
10. A signal processing method for a terahertz wireless communication system, characterized in that, include: An initial signal modulation symbol is generated based on an initial random bit sequence, and a digital intermediate frequency signal for the transmitting end is generated based on the initial signal modulation symbol. The digital intermediate frequency signal of the transmitting end is converted into an analog intermediate frequency signal of the transmitting end, and a terahertz signal is generated based on the local oscillator signal and the analog intermediate frequency signal of the transmitting end. Based on the terahertz signal and the local oscillator signal, an analog intermediate frequency (IF) signal for the receiving end is generated, and the analog IF signal for the receiving end is converted into a digital IF signal for the receiving end. The digital intermediate frequency signal at the receiving end is processed to obtain a digital baseband signal. The digital baseband signal is then resampled, retiming, linearly equalized, frequency offset estimated, and carrier phase recovered to obtain a carrier phase synchronization digital signal. Finally, the carrier phase synchronization digital signal is processed and demapped based on a sliding window bidirectional LSTM neural network to obtain a target random bit sequence.