A Deep Learning-Based Cross-Medium Visible Light Communication System and Method

Abstract

This invention provides a cross-medium visible light communication system and method based on deep learning. On one hand, it proposes an angle estimator and receiver movement strategy based on a convolutional neural network. The receiver can make movement decisions based on the angle estimation results, adaptively achieving link alignment of the transceiver in the cross-medium visible light communication system. On the other hand, this invention proposes a signal detection algorithm based on wind speed estimation and channel classification. After the system completes link alignment, this algorithm can estimate the wind speed and adaptively detect the received signal transmitted through the channel under different wind speed conditions. Furthermore, this invention proposes a loss function based on a modified mean square error criterion, which can effectively mitigate the adverse effects of wave-induced fading on signal transmission in the cross-medium visible light system communication channel, thereby reducing the system's bit error rate.

Description

Technical Field

[0001] This invention relates to the fields of deep learning and cross-medium visible light communication technology, and more specifically, to a deep learning-based cross-medium visible light communication system and method. Background Technology

[0002] In recent years, with the increasing demands for marine exploration and rescue, cross-medium communication has received widespread research attention. Visible Light Communication (VLC) systems have become a crucial technology for cross-medium communication due to their ability to transmit signals across the air-water interface and their numerous advantages, including high data rates and high security. However, current cross-medium VLC systems still face certain limitations and technical challenges related to light absorption, scattering, and dynamic wavefront refraction, requiring further in-depth research and solutions.

[0003] To improve the signal transmission performance of existing cross-medium VLC systems, researchers have mainly focused on optimizing the following two aspects:

[0004] On the one hand, from the perspective of link alignment (LA) optimization design, a wide beam can be used to expand the coverage of LEDs, or non-line-of-sight (NLOS) communication can be used to improve link reliability. However, the disadvantage of this optimization scheme is that the signal transmission distance is relatively limited. In addition, multiple-input multiple-output (MIMO) technology can be used to relax alignment requirements, or beam tracking technology can be used to maintain alignment in a dynamic environment, but the cost is increased equipment cost and complexity.

[0005] On the other hand, advanced transmission technologies are being considered to improve the reliability and throughput of cross-medium VLC. For example, Orthogonal Frequency Division Multiplexing (OFDM) technology can be combined to increase the transmission rate, or Adaptive Threshold (ATD) based demodulation techniques can be used to improve bit error rate performance. Adaptive Differential Pulse Position Modulation (ADPPM) technology can also be used to enhance system robustness. However, for optimization in this area, existing technologies have not fully considered how to adapt to dynamic wavefronts, or struggle to meet the demands of high transmission rates. Summary of the Invention

[0006] To overcome the shortcomings of existing cross-medium visible light communication systems, such as limited signal transmission distance, high equipment cost and complexity, difficulty in adapting to dynamic wavefronts, and difficulty in balancing high communication rates and bit error rates, this invention provides a deep learning-based cross-medium visible light communication system and method. Based on deep learning (DL) technology, it can adaptively achieve link alignment of transceivers in cross-medium VLC systems. After the system completes alignment, it achieves high-precision adaptive signal detection under different conditions, while effectively mitigating the adverse effects of wave-induced fading in cross-medium VLC channels on signal transmission, thereby reducing the system's bit error rate (BER).

[0007] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0008] A deep learning-based cross-medium visible light communication system includes: a transmitter, a transmission channel, and a receiver; the transmitter and receiver are respectively placed in different signal transmission media.

[0009] The receiving end includes: a signal receiving module, a data preprocessing module, a link alignment module, and a signal detection module;

[0010] The transmitting end is used to generate and transmit optical signals, and the optical signals are transmitted through the transmission channel and then reach the signal receiving module of the receiving end.

[0011] The data preprocessing module at the receiving end is used to preprocess the received optical signal to obtain pilot signal and data signal;

[0012] The link alignment module of the receiving end is equipped with an angle estimator based on a deep learning algorithm. The link alignment module is used to calculate the equivalent angle between the receiving end and the transmitting end based on the pilot signal, and adjust the position of the receiving end according to the equivalent angle until link alignment is achieved.

[0013] The signal detection module of the receiving end is equipped with a wind speed estimator based on a deep learning algorithm, a channel classifier, and several signal detectors. The signal detection module is used to estimate the wind speed based on the pilot signal and calculate the weight value corresponding to each signal detector after the link is aligned, and to detect the data signal in real time to realize cross-medium visible light communication.

[0014] Preferably, the transmitting end generates an optical signal through optical OFDM (Optical OFDM, OOFDM) modulation, wherein the optical OFDM modulation includes any one of asymmetrically clipped optical orthogonal frequency division multiplexing (ACO-OFDM) and direct current biased optical orthogonal frequency division multiplexing (DCO-OFDM).

[0015] Preferably, the equivalent angle is the angle between the line connecting the transmitter and receiver and the vertical direction;

[0016] The vertical distance between the receiver and the transmitter remains constant; the signal receiving plane of the receiver includes several concentric circles of different radii, each concentric circle representing an angle layer, and the equivalent angle between the receiver and the transmitter is the same at any position in each angle layer.

[0017] Preferably, the transmitter is mounted on an autonomous underwater vehicle (AUV), which is positioned underwater.

[0018] The receiving end is installed on an unmanned aerial vehicle (UAV) and the UAV is positioned in the airspace at a specific height above the water surface.

[0019] This invention also provides a deep learning-based cross-media visible light communication method, based on the aforementioned deep learning-based cross-media visible light communication system, comprising the following steps:

[0020] S1: The transmitter will transmit a bit vector. After mapping and optical modulation, a time-domain transmitted signal vector is generated. And transmit it to the receiving end through the transmission channel;

[0021] S2: The signal receiving module receives the time-domain transmitted signal vector. It then performs photoelectric conversion to acquire the time-domain received signal. ;

[0022] S3: The data preprocessing module processes the received signal in the time domain. Preprocessing is performed to obtain the received signal. The received signal Including pilot signals and data signals ;

[0023] S4: In the link alignment module, the angle estimator is based on the pilot signal. Calculate the equivalent angle between the receiver and the transmitter. And according to the equivalent angle Adjust the position of the receiving end until the link is aligned, then proceed to step S5;

[0024] S5: In the signal detection module, the wind speed estimator is based on the pilot signal. Estimated wind speed And input to the channel classifier, which is based on wind speed Calculate the weight value corresponding to each signal detector;

[0025] The signal detector detects data signals in real time. And calculate the soft bit vector based on the corresponding weight values. Cross-media visible light communication is achieved by recovering data bits through hard decision.

[0026] Preferably, in step S4, based on the equivalent angle Adjusting the position of the receiving end until link alignment is achieved includes:

[0027] Real-time based on the equivalent angle Determine the angle layer of the receiving end According to the angle layer Determine the next movement step size of the receiver;

[0028] Real-time based on the current equivalent angle Determine the next moving direction of the receiving end, and move the receiving end one step along the moving direction each time, continuously moving the position of the receiving end until the current equivalent angle is reached. Less than or equal to the preset alignment angle threshold The position of the receiving end is fixed to achieve link alignment.

[0029] Preferably, the angle layer The larger the radius, the larger the next movement step of the receiver;

[0030] The real-time calculation is based on the current equivalent angle. Determine the next moving direction of the receiving end, and move the receiving end one step along the moving direction each time. Continuously moving the position of the receiving end includes:

[0031] S41: The receiving end moves one step along a preset direction;

[0032] S42: Determine the current equivalent angle Is it less than or equal to the equivalent angle before the movement? If so, the receiving end keeps the preset direction unchanged, moves one step along the preset direction, and repeats step S42; otherwise, execute step S43.

[0033] S43: The receiving end returns to its previous position and moves one step along the vertical direction of any side of the preset direction;

[0034] After moving perpendicularly to either side of the preset direction, determine the current equivalent angle. Is it less than or equal to the equivalent angle before the movement? If so, then the vertical direction is taken as the new preset direction and step S41 is executed again; otherwise, the opposite direction of the vertical direction is taken as the new preset direction and step S41 is executed again.

[0035] Preferably, in step S4, the angle estimator is specifically a trained first CNN network; the input to the first CNN network is the pilot signal. The output is the equivalent angle. The loss function used to supervise the training of the first CNN network is the cross-entropy loss function.

[0036] In step S5, the wind speed estimator is specifically a trained second CNN network; the input to the second CNN network is the pilot signal. The output is wind speed. The estimated value The loss function used to supervise the training of the second CNN network is the cross-entropy loss function.

[0037] Each of the signal detectors is a trained third CNN network; the input of the third CNN network is the data signal. The output is a soft bit vector. The loss function used to supervise the training of the third CNN network is the mean squared error loss function.

[0038] The channel classifier is specifically a trained DNN network, and the input of the DNN network is wind speed. The estimated value The output is the weight value corresponding to each signal detector; the loss function used to supervise the training of the DNN network is a loss function based on the CMSE criterion.

[0039] Preferably, the loss function formula based on the CMSE criterion is as follows:

[0040]

[0041] in, For the correction term coefficient; It is the expected value function; This represents the weight value corresponding to the i-th signal detector; This represents the number of signal detectors.

[0042] Preferably, in step S3, the data preprocessing module processes the time-domain received signal. The received signal is preprocessed by sequentially performing analog-to-digital conversion, serial-to-parallel conversion, removal of cyclic prefixes, and fast Fourier transform. .

[0043] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0044] This invention provides a cross-media visible light communication system and method based on deep learning. On one hand, it proposes an angle estimator and receiver movement strategy based on a convolutional neural network (CNN). The receiver can make movement decisions based on the angle estimation results, and adaptively realize the link alignment of the transceiver in the cross-media visible light communication system.

[0045] On the other hand, this invention proposes a signal detection (SD) algorithm based on wind-speed estimation (WE) and channel classification (CC). After the system completes link alignment, this algorithm can estimate the wind speed and adaptively detect the received signal transmitted through the channel under different wind speed conditions. In this algorithm, this invention also proposes a loss function based on the correction of mean square error (CMSE) criterion, which can effectively mitigate the adverse effects of wave-induced fading on signal transmission in cross-medium visible light system communication channels, thereby reducing the bit error rate of the system. Attached Figure Description

[0046] Figure 1 This is a structural diagram of a cross-media visible light communication system based on deep learning provided in Example 1.

[0047] Figure 2 This is a schematic diagram of the W2A-VLC system communication scenario provided in Example 1.

[0048] Figure 3 This is a schematic diagram of the signal receiving plane of the receiver provided in Embodiment 1.

[0049] Figure 4This is a flowchart of a cross-media visible light communication method based on deep learning provided in Example 2.

[0050] Figure 5 This is a flowchart of the receiving end mobile strategy processing provided in Example 2.

[0051] Figure 6 This is a graph showing the accuracy performance of the CNN angle estimator provided in Example 3.

[0052] Figure 7 This is an example diagram of the UAV movement trajectory provided in Example 3.

[0053] Figure 8 The bit error rate performance graph of the CNN signal detector provided in Example 3.

[0054] Figure 9 The graph shows the accuracy performance of the CNN wind speed estimator provided in Example 3.

[0055] Figure 10 As provided in Example 3 Bit error rate performance of the WE-CC-SD algorithm at m / s.

[0056] Figure 11 As provided in Example 3 Bit error rate performance of the WE-CC-SD algorithm at m / s. Detailed Implementation

[0057] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0058] To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions;

[0059] It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.

[0060] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0061] Example 1

[0062] like Figure 1 As shown, this embodiment provides a cross-media visible light communication system based on deep learning, including: a transmitter, a transmission channel, and a receiver; the transmitter and receiver are respectively placed in different signal transmission media;

[0063] The receiving end includes: a signal receiving module, a data preprocessing module, a link alignment module, and a signal detection module;

[0064] The transmitting end is used to generate and transmit optical signals, and the optical signals are transmitted through the transmission channel and then reach the signal receiving module of the receiving end.

[0065] The data preprocessing module at the receiving end is used to preprocess the received optical signal to obtain pilot signal and data signal;

[0066] The link alignment module of the receiving end is equipped with an angle estimator based on a deep learning algorithm. The link alignment module is used to calculate the equivalent angle between the receiving end and the transmitting end based on the pilot signal, and adjust the position of the receiving end according to the equivalent angle until link alignment is achieved.

[0067] The signal detection module of the receiving end is equipped with a wind speed estimator based on a deep learning algorithm, a channel classifier, and several signal detectors. The signal detection module is used to estimate the wind speed based on the pilot signal and calculate the weight value corresponding to each signal detector after the link is aligned, and to detect the data signal in real time to realize cross-medium visible light communication.

[0068] The transmitter generates an optical signal through optical OFDM modulation, which includes any one of asymmetric limiting optical-orthogonal frequency division multiplexing and DC biased optical-orthogonal frequency division multiplexing.

[0069] The equivalent angle is the angle between the line connecting the transmitter and receiver and the vertical direction.

[0070] The vertical distance between the receiver and the transmitter remains constant; the signal receiving plane of the receiver includes several concentric circles of different radii, each concentric circle represents an angle layer, and the equivalent angle between the receiver and the transmitter is the same at any position in each angle layer;

[0071] The transmitter is mounted on an autonomous underwater vehicle (AUV), which is positioned underwater.

[0072] The receiving end is mounted on a UAV, which is positioned in an airspace at a specific height above the water surface.

[0073] In the specific implementation process, this embodiment proposes a W2A-VLC system based on deep learning for the Water-to-Air (W2A) cross-medium visible light communication scenario, and solves the link alignment problem and adaptive signal detection problem. In this embodiment, the designed system can be divided into three parts: AUV transmitter, W2A-VLC channel and UAV receiver.

[0074] The communication scenario diagram of this embodiment is as follows: Figure 2As shown, the UAV flies in the airspace at a certain height above the water surface and communicates with the AUV transmitter through the W2A-VLC channel; Figure 2 middle, It is the distance of the UAV from the horizontal plane. It is the distance of the AUV from the horizontal plane, equivalent angle. It is the angle between the line connecting the UAV and AUV and the vertical direction; such as Figure 3 As shown, this is the flight plane of the UAV receiver. The flight plane consists of several concentric circles with different radii. Each concentric circle represents an angular layer, and each angular layer represents... The set of equal positions is defined as the angle layer. ,in The number of angle layers;

[0075] exist Figure 1 middle, This represents the pilot signal vector received during the LA process. This refers to the pilot signal received during data transmission. Indicates data signal, This represents the estimated equivalent angle. Indicates wind speed The estimated value, express Input signal detector i The output soft bit vector, express The corresponding weight values; the communication process of this system is as follows:

[0076] The AUV transmitter will transmit a bit vector. After mapping and optical OFDM modulation, the time-domain transmitted signal vector is obtained. ,and After transmission through the W2A-VLC channel, the signal reaches the receiving end for subsequent processing. This system can adopt a variety of common OOFDM schemes, such as asymmetric limiting optical-orthogonal frequency division multiplexing, DC biased optical-orthogonal frequency division multiplexing, etc.

[0077] At the receiving end, the signal receiving module receives the signal and performs photoelectric conversion to obtain the time-domain received signal. ;

[0078] The data preprocessing module processes the received signal in the time domain. The received signal is obtained by performing analog-to-digital conversion, serial-to-parallel conversion, removal of cyclic prefix, and fast Fourier transform. This includes pilot signals. and data signals ;

[0079] Link alignment module according to Determine if the UAV is aligned and make a movement decision until the alignment conditions are met. Then, the UAV hovers at the alignment position and the signal detection module is activated.

[0080] Signal detection module based on Perform signal detection to obtain the estimated soft bit vector of the transmitted bit vector. Then, the data bits are recovered through hard decision to complete the detection of the W2A-VLC signal;

[0081] This system performs link alignment based on angle estimation and adaptive channel selection and signal detection based on wind speed estimation. Compared with existing technologies, this system can achieve adaptive and continuous alignment of the UAV receiver with the AUV transmitter. It can also achieve adaptive wave mitigation and low bit error rate signal detection under dynamic channels.

[0082] Example 2

[0083] like Figure 4 As shown, this embodiment provides a deep learning-based cross-media visible light communication method, based on the deep learning-based cross-media visible light communication system described in Embodiment 1, including the following steps:

[0084] S1: The transmitter will transmit a bit vector. After mapping and optical modulation, a time-domain transmitted signal vector is generated. And transmit it to the receiving end through the transmission channel;

[0085] S2: The signal receiving module receives the time-domain transmitted signal vector. It then performs photoelectric conversion to acquire the time-domain received signal. ;

[0086] S3: The data preprocessing module processes the received signal in the time domain. Preprocessing is performed to obtain the received signal. The received signal Including pilot signals and data signals ;

[0087] S4: In the link alignment module, the angle estimator is based on the pilot signal. Calculate the equivalent angle between the receiver and the transmitter. And according to the equivalent angle Adjust the position of the receiving end until the link is aligned, then proceed to step S5;

[0088] S5: In the signal detection module, the wind speed estimator is based on the pilot signal. Estimated wind speed And input to the channel classifier, which is based on wind speed Calculate the weight value corresponding to each signal detector;

[0089] The signal detector detects data signals in real time. And calculate the soft bit vector based on the corresponding weight values. Cross-media visible light communication is achieved by recovering data bits through hard decision;

[0090] In step S4, based on the equivalent angle Adjusting the position of the receiving end until link alignment is achieved includes:

[0091] Real-time based on the equivalent angle Determine the angle layer of the receiving end According to the angle layer Determine the next movement step size of the receiver;

[0092] Real-time based on the current equivalent angle Determine the next moving direction of the receiving end, and move the receiving end one step along the moving direction each time, continuously moving the position of the receiving end until the current equivalent angle is reached. Less than or equal to the preset alignment angle threshold The position of the receiving end is fixed to achieve link alignment;

[0093] The angle layer The larger the radius, the larger the next movement step of the receiver;

[0094] The real-time calculation is based on the current equivalent angle. Determine the next moving direction of the receiving end, and move the receiving end one step along the moving direction each time. Continuously moving the position of the receiving end includes:

[0095] S41: The receiving end moves one step along a preset direction;

[0096] S42: Determine the current equivalent angle Is it less than or equal to the equivalent angle before the movement? If so, the receiving end keeps the preset direction unchanged, moves one step along the preset direction, and repeats step S42; otherwise, execute step S43.

[0097] S43: The receiving end returns to its previous position and moves one step along the vertical direction of any side of the preset direction;

[0098] After moving perpendicularly to either side of the preset direction, determine the current equivalent angle. Is it less than or equal to the equivalent angle before the movement? If so, then the vertical direction is taken as the new preset direction and step S41 is executed again; otherwise, the opposite direction of the vertical direction is taken as the new preset direction and step S41 is executed again.

[0099] In step S4, the angle estimator is specifically a trained first CNN network; the input to the first CNN network is the pilot signal. The output is the equivalent angle. The loss function used to supervise the training of the first CNN network is the cross-entropy loss function.

[0100] In step S5, the wind speed estimator is specifically a trained second CNN network; the input to the second CNN network is the pilot signal. The output is wind speed. The estimated value The loss function used to supervise the training of the second CNN network is the cross-entropy loss function.

[0101] Each of the signal detectors is a trained third CNN network; the input of the third CNN network is the data signal. The output is a soft bit vector. The loss function used to supervise the training of the third CNN network is the mean squared error loss function.

[0102] The channel classifier is specifically a trained DNN network, and the input of the DNN network is wind speed. The estimated value The output is the weight value corresponding to each signal detector; the loss function used to supervise the training of the DNN network is a loss function based on the CMSE criterion;

[0103] The specific formula for the loss function based on the CMSE criterion is as follows:

[0104]

[0105] in, For the correction term coefficient; It is the expected value function; This represents the weight value corresponding to the i-th signal detector; The number of signal detectors;

[0106] In step S3, the data preprocessing module processes the time-domain received signal. The received signal is preprocessed by sequentially performing analog-to-digital conversion, serial-to-parallel conversion, removal of cyclic prefixes, and fast Fourier transform. .

[0107] In the specific implementation process, the transmitting end will first transmit the bit vector. After mapping and optical modulation, a time-domain transmitted signal vector is generated. And it is sent to the receiving end through the transmission channel;

[0108] Next, the signal receiving module receives the time-domain transmitted signal vector. It then performs photoelectric conversion to acquire the time-domain received signal. ;

[0109] The data preprocessing module processes the received signal in the time domain. The received signal is preprocessed by sequentially performing analog-to-digital conversion, serial-to-parallel conversion, removal of cyclic prefixes, and fast Fourier transform. Receive signal Including pilot signals and data signals ;

[0110] Then, link alignment is performed using the link alignment module. Figure 2 In this configuration, the AUV transmitter transmits signals vertically, while the UAV receiver operates at a constant height. On the plane, based on the angle estimation results, obtain the current angle layer. Then, a corresponding movement strategy is adopted, moving an appropriate step length each time along the "zigzag spiral" trajectory, and finally reaching the alignment point;

[0111] Specifically, the angle estimator is based on the pilot signal. Calculate the equivalent angle between the receiver and the transmitter. The angle estimator is a pre-trained CNN network, and the input is the pilot signal. The output is the equivalent angle. ;

[0112] Then, based on the equivalent angle in real time Determine the angle layer of the receiving end According to the angle layer Determine the next movement step size of the receiver; Angular layer The larger the radius, the larger the next movement step of the receiver (i.e., the outer angle layer corresponds to a longer step, and the inner angle layer corresponds to a shorter step).

[0113] Real-time based on the current equivalent angle Determine the next direction of movement for the receiver. Each time, move the receiver one step along the direction of movement, continuously moving the receiver's position until the current equivalent angle is reached. Less than or equal to the preset alignment angle threshold Fix the receiver position to achieve link alignment;

[0114] like Figure 5As shown, this is based on the current equivalent angle. The flowchart for determining the mobility strategy shows that the UAV adjusts the position of the receiving end according to the mobility strategy. The specific steps are as follows:

[0115] S41: The receiving end moves one step along a preset direction (random direction);

[0116] S42: Determine the current equivalent angle Is it less than or equal to the equivalent angle before the movement? If so, the receiving end keeps the preset direction unchanged, moves one step along the preset direction, and repeats step S42; otherwise, execute step S43.

[0117] S43: The receiving end returns to its previous position and moves one step along the vertical direction of any side of the preset direction;

[0118] After moving perpendicularly to either side of the preset direction, determine the current equivalent angle. Is it less than or equal to the equivalent angle before the movement? If so, then the vertical direction is taken as the new preset direction and step S41 is executed again; otherwise, the opposite direction of the vertical direction is taken as the new preset direction and step S41 is executed again.

[0119] It is worth mentioning that after the link is aligned, the wind speed is estimated. This information can be used as supplementary information to help the UAV better adjust its control parameters and maintain a hovering state at the alignment point, which is beneficial for maintaining persistent link alignment. For example, the UAV can adjust the rotation speed of the corresponding rotor motor based on the estimated wind speed to compensate for displacement and attitude changes caused by the wind. In addition, hovering control can also be achieved based on the special technologies of UAV manufacturers to maintain a hovering state at the alignment point. If the UAV deviates from its position due to environmental factors, the above link alignment process can be repeated to re-acquire the alignment point. Since UAV hovering control technology belongs to other professional fields such as automatic control and mechanics, it is not within the scope of protection of this invention and will not be described in detail here.

[0120] After link alignment is completed, the system can perform data communication, at which point the signal detection module is activated;

[0121] exist Time-domain received signal at time for:

[0122]

[0123] in, Transmitted signals in the time domain; * represents the impulse response; * represents temporal convolution; For noise;

[0124] By wind speed The determined Pearson-Mills (PM) wave power spectrum is as follows:

[0125]

[0126] in: It is a constant factor. It is gravitational acceleration. It is the angular frequency; according to the above formula, different Different wavefronts are formed by the values, which affect the channel through refraction on the sea surface. This causes wave-induced fading, which in turn leads to a decrease in the system's detection performance;

[0127] To achieve wave mitigation, this method proposes a signal detection module consisting of three parts: a wind speed estimator composed of a CNN network, several signal detectors, and a channel classifier composed of a deep neural network (DNN). The CNN signal detectors, after training, can detect the received data signals... Obtain the estimated soft bit vector To improve the system's adaptability to corresponding channels under untrained wind speed conditions, this method targets... Training at different wind speeds A CNN signal detector was developed for subsequent signal detection under various unknown wind speeds.

[0128] In the absence of wind speed, enter The output is obtained by using a CNN signal detector. The CNN wind speed estimator then uses the received pilot signals. Estimate the wind speed corresponding to the channel at this time. and the estimated value Inputting the DNN channel classifier yields the result used for A single combined weight vector of a CNN detector ,but It can be obtained from the following formula:

[0129]

[0130] Furthermore, according to the formula for wave power spectrum, in order to ensure that the decision threshold for signal detection remains unchanged, the following conditions must be met: Therefore, this method designs a loss function based on the CMSE criterion for DNN channel classifiers to help the system achieve better bit error rate performance, as shown in the following formula:

[0131]

[0132] in, For the correction term coefficient; It is the expected value function; This represents the weight value corresponding to the i-th signal detector; The number of signal detectors;

[0133] After link alignment is completed, signal detection is performed using a signal detection module, where the wind speed estimator is based on the pilot signal. Estimated wind speed And input it into the channel classifier, which classifies it according to wind speed. Calculate the weight value corresponding to each signal detector;

[0134] Signal detectors detect data signals in real time. And calculate the soft bit vector based on the corresponding weight values. Cross-media visible light communication is achieved by recovering data bits through hard decision;

[0135] This method proposes an angle estimator and receiver movement strategy based on a convolutional neural network. The receiver can make movement decisions based on the angle estimation results, adaptively achieving link alignment of the transceiver in a cross-medium visible light communication system. In addition, this method proposes a signal detection algorithm based on wind speed estimation and channel classification. After the system completes link alignment, this algorithm can estimate the wind speed and adaptively detect the received signal transmitted through the channel under different wind speed conditions. In this algorithm, this method also proposes a loss function based on a modified mean square error criterion, which can effectively mitigate the adverse effects of wave-induced fading on signal transmission in cross-medium visible light communication channels, thereby reducing the bit error rate of the system.

[0136] Example 3

[0137] This embodiment provides a simulation experiment to verify the effectiveness of a deep learning-based cross-media visible light communication method proposed in Embodiment 2.

[0138] In this specific implementation, DCO-OFDM is selected as the modulation scheme, with a DC bias of 13dB, 64 subcarriers, and QPSK modulation. Each signal frame contains 10 data OFDM symbols and 1 pilot OFDM symbol. The W2A-VLC channel is obtained based on a Monte Carlo simulation model, and the distance of the UAV from the horizontal plane is... m, distance of AUV from the horizontal plane m, the diameter of PD (Photo Diode) is m, LED light source wavelength is nm, LED half-power angle is The simulated photon number is PD half field of view is The simulation uses Jerlov I water bodies from the Jerlov water classification system, with absorption and scattering coefficients... , for ;

[0139] In the simulation of the link alignment module, the network structure and main parameters of the CNN angle estimator are shown in Table 1, where: The number of convolutional layers. The number of convolution kernels, For the convolution kernel dimension, The convolution stride is... The number of fully connected layers. This represents the number of neurons in the fully connected layer.

[0140] Table 1: Main parameters of the CNN angle estimator

[0141]

[0142] This embodiment is set Each angle layer, namely ,in: During training, wind speed is taken. m / s, randomly select 4 positions on each of the above angle layers and obtain the value at each position. One received signal Obtained from all locations This constitutes the training set; in particular, due to The mean square error of the channel sample values ​​within the internal distribution is less than Therefore, choose As the alignment angle threshold; since the receiver planar area corresponding to the same angle varies... , The values ​​of are different, so they should be determined according to . , Choose an appropriate value Value; furthermore, the training strategy adopted in this embodiment is to respectively in Each training session consists of 100 epochs.

[0143] During the test, four wind speeds were used. Under the condition of m / s, Inside, and between, and between, and Four sub-angle layers are randomly selected from each of the two layers, and data is collected from each sub-angle layer. At random positions The data was used as a test set.

[0144] Finally, the input vector dimension of the CNN network is set to... The training set size is There are n input vectors, and the test set size is [size missing]. Input vectors; the accuracy performance of the CNN angle estimator is as follows: Figure 6 As shown;

[0145] like Figure 6 As shown, the angle estimator proposed in this invention performs well under high SNR conditions (such as...). It can achieve 100% accuracy; at the same time, The network trained at m / s can also adapt to various untrained wind speed conditions, which proves the robustness of the scheme proposed in Example 2.

[0146] Furthermore, Figure 7 This provides an example trajectory diagram of a UAV moving in the receiver plane, where the angles from the inside to the outside of the angle layer are respectively... The numbers 1 to 10 represent the order of the UAV's movement trajectory; the thick arrows represent the direction of movement of the UAV; and the thin arrows represent each segment of the historical movement trajectory. Figure 7 As can be seen, the UAV can reach the alignment point after several movements, proving the effectiveness of the solution proposed in Example 2;

[0147] After alignment is completed, the W2A-VLC system can perform data communication; in the simulation of the WE-CC-SD algorithm, the network structure and main parameters are shown in Table 2;

[0148] Table 2: Main parameters of WE-CC-SD

[0149]

[0150] The simulation of the WE-CC-SD algorithm consists of the following three parts:

[0151] 1) First part: Using data signals The CNN signal detector was trained and tested using the same training strategy as described above, with specific parameters shown in Table 2. The output of the trained network is the estimated soft bit vector. In this embodiment, the following options are selected respectively. m / s (still water) and Data generated at a channel speed of m / s was used for training and testing of a CNN signal detector; the bit error rate performance of the two CNN signal detectors was as follows: Figure 8 As shown;

[0152] Depend on Figure 8It can be seen that the performance of the CNN signal detector trained alone is close to that of the ideal scheme under the assumption of perfect Channel State Information (CSI) conditions, and it outperforms the least squares (LS) estimation and linear minimum mean square error (LMMSE) estimation schemes under the condition of wave-induced fading.

[0153] 2) Second part: The received pilot signal Used for training and testing of a CNN wind speed estimator; respectively generated in under wind speed conditions of m / s The actual wind speed will be used as input data for training. One-hot encoding is used as the ground truth to train the network output corresponding estimated wind speed. The loss function is the cross-entropy between the two; the accuracy performance of the CNN wind speed estimator can be found in [link to relevant documentation]. Figure 9 It can be seen that the CNN wind speed estimator has a high accuracy under high SNR conditions;

[0154] 3) Part Three: Training the DNN channel classifier and testing the overall performance of the WE-CC-SD algorithm; Generated at a wind speed of m / s The signals were input into the signal detector trained in the first part, and 11 sets of signals corresponding to the 11 wind speeds were obtained. Then 11 groups and corresponding wind speed The one-hot encoding is used as the input to the channel classifier; then, the network is trained using the CMSE loss function proposed in Example 2 to obtain the corresponding combined weights. ; Figure 10 and Figure 11 Tested separately m / s and Bit error rate performance of the WE-CC-SD algorithm at m / s;

[0155] Depend on Figure 10 and Figure 11 As can be seen, the WE-CC-SD algorithm consistently outperforms traditional schemes in terms of bit error rate, verifying that the algorithm can enable the system to operate under unknown wind speeds. To achieve better bit error rate performance under certain conditions.

[0156] The same or similar labels correspond to the same or similar parts;

[0157] The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.

[0158] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A deep learning based cross-media visible light communication system, characterized in that, The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device.

2. The deep learning based cross-medium visible light communication system of claim 1, wherein, The application relates to a cross-medium visible light communication method and device. 3.The cross-medium visible light communication system based on deep learning according to any one of claims 1-2, characterized in that, The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device.

4. A method for cross-medium visible light communication based on deep learning, based on the cross-medium visible light communication system based on deep learning in any one of claims 1-3, characterized in that, The application relates to a cross-medium visible light communication method and device. S1: the transmitting end transmits a bit vector After mapping and light modulation, a time-domain transmitting signal vector is generated and is transmitted to the receiving end through the transmission channel; S2: the signal receiving module receives a time-domain transmitting signal vector and performs photoelectric conversion to obtain a time-domain receiving signal ; S3: the data preprocessing module pre-processes the time-domain received signal to obtain a received signal , wherein the received signal includes a pilot signal and a data signal ; S4: In the link alignment module, an angle estimator estimates the angle between the receiving end and the transmitting end according to the pilot signals calculates the equivalent angle between the receiving end and the transmitting end and adjusts the position of the receiving end according to the equivalent angle until the link alignment is achieved, step S5 is executed; S5: In the signal detection module, the wind speed estimator is based on the pilot signal. Estimated wind speed And input to the channel classifier, which is based on wind speed Calculate the weight value corresponding to each signal detector; The signal detector detects the data signal in real time and calculates a soft bit vector according to the corresponding weight value and realizes cross-medium visible light communication through hard decision recovery of data bits.

5. The method of claim 4, wherein, In the step S4, the equivalent angle is calculated according to the formula Adjusting the position of the receiving end until the link alignment is achieved includes: determining the angle layer where the receiving end is located determining the angle layer where the receiving end is located determining the angle layer where the receiving end is located determining the moving step of the receiving end next time determining the next moving direction of the receiving end each time moving the receiving end by one step along the moving direction, and constantly moving the position of the receiving end until the current equivalent angle determining the next moving direction of the receiving end each time moving the receiving end by one step along the moving direction, and constantly moving the position of the receiving end until the current equivalent angle less than or equal to a preset alignment angle threshold fixing the position of the receiving end to realize link alignment.

6. The method of claim 5, wherein the method is based on deep learning. The angle layer The greater the radius, the greater the next movement step of the receiving end. The real-time according to the current equivalent angle Determining the next moving direction of the receiving end, moving the receiving end by one step along the moving direction each time, and constantly moving the position of the receiving end include: The application relates to a cross-medium visible light communication method and device. S42: judging whether the current equivalent angle is less than or equal to the equivalent angle before the movement S43: judging whether the current equivalent angle is less than or equal to the equivalent angle before the movement If yes, the receiving end keeps the preset direction unchanged, continues to move one step along the preset direction, and re-executes step S42; otherwise, step S43 is executed. The application relates to a cross-medium visible light communication method and device. After moving in the vertical direction on either side of the preset direction, the current equivalent angle is determined whether it is less than or equal to the equivalent angle before the movement If yes, the vertical direction is taken as the new preset direction, and step S41 is re-executed; otherwise, the opposite direction of the vertical direction is taken as the new preset direction, and step S41 is re-executed.

7. The method of claim 4, wherein the method is based on deep learning. The angle estimator in the step S4 is specifically a trained first CNN network; an input of the first CNN network is the pilot signal , and an output is the equivalent angle ; a loss function for supervising the training of the first CNN network is a cross-entropy loss function; The wind speed estimator in the step S5 is specifically a trained second CNN network; an input of the second CNN network is the pilot signal , and an output is an estimated value of the wind speed ; a loss function used for supervising the training of the second CNN network is a cross-entropy loss function . Each of the signal detectors is a trained third CNN network; an input of the third CNN network is the data signal , and an output is a soft bit vector ; a loss function used for supervising training of the third CNN network is a mean square error loss function; The channel classifier is specifically a trained DNN network, an input of the DNN network is an estimated value of the wind speed , and an output is a weight value corresponding to each signal detector; a loss function used for supervising training of the DNN network is a loss function based on a CMSE criterion.​ 8. The method of claim 7, wherein, The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to a cross-medium visible light communication method and device. The application relates to wherein, is a correction term coefficient; is an expected value function; is a weight value corresponding to the i-th signal detector; is the number of signal detectors.

9. The method of claim 4-8, wherein, In the step S3, the data preprocessing module pre-processes the time-domain received signal Analog-to-digital conversion, serial-to-parallel conversion, cyclic prefix removal and fast Fourier transform are sequentially performed to complete the preprocessing and obtain the received signal .

Images