A spatial light cooperation transmission method based on elastic light splitting

Abstract

The application relates to a spatial light cooperation transmission method based on elastic splitting, and belongs to the field of optical communication. In view of the problem that a fixed optical splitter is used in a traditional cooperation transmission system and is not conducive to full utilization of optical power, a spatial light cooperation transmission method based on elastic splitting is provided. The main idea is to design a cascaded optical splitter structure based on a phase shifter and a coupler, and according to different spatial channel states, flexible power distribution between different spatial light links can be realized. Based on the elastic optical splitter, a deep reinforcement learning algorithm for spatial light relay selection and power distribution is further designed, and by optimizing relay selection and power distribution, the transmission performance of the spatial light cooperation transmission system under different channel conditions is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of optical communication, and relates to a spatial light cooperative transmission method based on elastic splitting. BACKGROUND

[0002] With the continuous development of modern information society, people's demand for wireless communication is getting larger and larger, and the requirement is getting higher and higher. The traditional communication mode has been unable to meet people's requirements for information transmission. Therefore, free space optical (FSO) communication technology emerges as the times require. FSO communication technology uses laser as transmission carrier and free space as transmission medium, and can realize high-capacity and high-speed information transmission without spectrum license. At the same time, the infrastructure of FSO communication does not need to deploy cable, and the equipment installation is flexible, which can provide high-speed communication connection for remote mountainous areas, islands, space and other areas where optical fiber cannot reach. Therefore, FSO communication technology is listed as one of the strategic key technologies for developing space-ground-sea integrated communication network construction in China.

[0003] However, the transmission of laser signal in space is easily affected by atmospheric reflection and scattering, causing loss. At the same time, the non-uniformity of temperature change and air pressure in the atmospheric channel will cause the refractive index fluctuation, i.e. atmospheric turbulence. Therefore, the communication quality of FSO is obviously affected by the atmospheric channel condition. In order to suppress the light intensity fading caused by atmospheric turbulence and loss and expand the coverage range of optical communication system, relay cooperative technology as an effective diversity technology is widely researched and applied. The traditional spatial light cooperative transmission system realizes the broadcast transmission of optical signal to different relays through a fixed optical splitter, and then receives and combines the optical signal through spatial diversity. However, this way will cause a large amount of optical power waste under the condition that the relay link channel is not ideal, which is not conducive to the improvement of system bit error rate performance. In order to further improve the optical power utilization rate and reduce the bit error rate of system transmission, the present application proposes a FSO cooperative transmission method based on elastic splitting. In view of the time-varying nature of atmospheric channel state, the present application designs a deep reinforcement learning algorithm which can adaptively decide the relay selection and optical power allocation according to the channel condition, and then distributes the input optical signal to the relay link as needed through an elastic optical splitter. The present application can adapt to the time-varying nature of atmospheric channel, flexibly adjust the relay set and the optical power output distributed to different relay links, so as to improve the transmission performance of the system.

[0004] After searching, there are few reports on the invention of elastic optical splitters. The patent with application publication number CN114966990A designs an adjustable optical splitter, which is characterized by controlling the number of output ports of the adjustable optical splitting device and adjusting the optical power of the output ports. The adjustable optical splitting device includes a Y-type coupler, an optical switch, and a 1:N star coupler, which can realize three cases of optical power adjustment: 1) the optical power of two output ports is 50%, or 2) the optical power of output port 1 is 100% and the optical power of port 2 is 0%, or 3) the optical power of output port 2 is 100% and the optical power of output port 1 is 0%. At the same time, the optical switch is used to control the number of output ports, and two optical switches are used in the patent description, which can realize the optical power of two output ports at most.

[0005] The present application adopts the cascading mode of phase shifter and coupler unit structure to realize the functions of arbitrary selection of the number of optical splitting output ports and arbitrary proportional adjustment of the output optical power. The optical splitting output is not limited to the limited optical splitting ratio of 100%:0%, 0%:100%, and 50%:50% of the existing adjustable optical splitter, and more accurate optical power distribution under the time-varying state of the atmospheric channel can be realized, effectively improving the flexibility and transmission efficiency of the system. According to different channel conditions, the elastic optical splitter designed by the present application can flexibly adjust the number of output ports and the optical power, and is no longer limited to the equal division of optical power or the output of all power to one port, but can be flexibly adjusted according to the actual situation. Therefore, the advantage of the present application is that the elastic optical splitter is applied to the FSO cooperative transmission system, the number of output ports can be selected according to different channel conditions, and the optical power in any proportion can be flexibly adjusted according to the channel conditions, so that the system achieves good communication quality and improves the anti-atmospheric loss and anti-turbulence ability of the system. SUMMARY

[0006] Therefore, the purpose of the present application is to provide a spatial light cooperative transmission method based on elastic optical splitting.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] A spatial light cooperative transmission method based on elastic optical splitting, the technical solutions of the present application are as follows:

[0009] The system sending end adopts an elastic optical splitter based on the cascading of phase shifters and couplers, which allows the input power to be flexibly distributed in any proportion among different output ports of the optical splitter.

[0010] Atmospheric channel adaptive relay link selection and optical power distribution, using a deep reinforcement learning algorithm, without channel data set, with the goal of minimizing the system bit error rate, dynamically selecting spatial light relay links according to different atmospheric channel states, and flexibly distributing the optical power of each link.

[0011] Furthermore, the specific steps of the atmospheric channel adaptive space optical relay link selection and optical power allocation include: 1) Under the channel conditions of time-varying turbulence intensity and atmospheric loss, the DQN (Deep Q Network) algorithm is used to continuously learn and fit the optimal action value function, predict and select the set of space optical relay links with the highest Q value, denoted as action. 2) Based on the time-varying state of the atmospheric channel, the Actor network simulation policy function in the DDPG (Deep Deterministic Policy Gradient) algorithm is used to predict the power allocation scheme of the selected space optical relay link, denoted as the action. Using the Critic network to simulate the action value function to calculate the action of the relay link power allocation scheme. The Q-value. The quality of the Actor network's predicted action is judged based on the Q-value; therefore, the objective of the policy function is to determine the action that maximizes the Q-value; 3) Selecting actions for relay links. and power distribution action Joint normalization is performed so that optical power is allocated to selected relay links in different proportions, while the optical power of unselected relay links is 0.

[0012] Define the Q-value objective function for space optical relay link selection and power allocation:

[0013]

[0014] in for The reward value at any moment, For discount rate, A value function model fitted to a neural network. for Q value at time t, for Channel state at any given time. for The set of actions under the current channel state. This represents the network weights. Using an atmospheric channel as the environment for deep reinforcement learning, the reward function is defined as follows:

[0015]

[0016] in BER (Bit Error Ratio) is the bit error rate, when BER Less than 10 -3 The reward is positive, and BERThe smaller the value, the higher the reward. After sending the relay link selection and optical power allocation scheme to the environment, the reward value and channel status are fed back in real time to update the objective function. Define the loss function:

[0017]

[0018] in for Q value at time t, for Channel state at any given time. for Actions under the channel conditions at any given time. Gradient descent is used to continuously optimize the neural network to minimize the loss function and maximize the Q-value until the reward converges, i.e., the cumulative expected reward is maximized. This allows for flexible adjustment of relay links and power allocation based on channel conditions, improving transmission stability and reliability. Detailed steps for adaptively adjusting relay link selection and optical power allocation using deep reinforcement learning algorithms are as follows: Figure 1 .

[0019] Furthermore, the steps of the designed flexible beam splitter structure specifically include: 1) Encoding the relay link selection and optical power allocation scheme and sending it to an electrical signal amplitude controller to generate a corresponding electrical signal to drive the phase shifter; 2) Splitting the input optical signal into two signals using a 1:1 beam splitter, and inputting one of the optical signals into the phase shifter, achieving phase shift of the optical signal under the drive of the electrical signal input in step 1; 3) Inputting the two optical signals with a phase difference into an X-coupler, and adjusting the phase difference to achieve different proportions of optical power output at the two output ports; 4) Arbitrarily modifying the unit structure obtained in steps 1, 2, and 3. N Class alliance, can achieve a 1:2 ratio. N A flexible beam splitter with a specific structure achieves 2D beam splitting by inputting different driving voltages to the phase shifter array. N The flexible beam splitter structure in this invention allows for arbitrary proportions of optical power output from each output port. Unlike the fixed beam splitters used in traditional space optical cooperative transmission systems, the flexible beam splitter structure in this invention can improve the optical power utilization of the space optical cooperative transmission system by flexibly selecting any number of output ports and flexibly adjusting the optical power of the output ports.

[0020] The adaptive spatial optical relay link selection, optical power allocation, and flexible beam splitter cooperative transmission method for atmospheric channels specifically includes: under different FSO channel states, after selecting the relay link and output optical power using a deep reinforcement learning algorithm, the relevant information is sent to an electrical signal amplitude controller to generate a corresponding electrical signal to drive a phase shifter. The phase shift of the optical signal can be achieved using a phase modulator as follows:

[0021]

[0022] In the formula is the output optical signal electric field, is the input optical signal electric field, is the phase shift, is the modulation electric signal.

[0023] The setting , by real-time control of the voltage value of the electric signal , the arbitrary phase shift of the input optical signal can be realized Then input two optical signals with phase difference and equal power into the coupler. According to the transmission function of the coupler:

[0024]

[0025] In the formula and are the input optical power of the coupler, and , and are the output optical power of the coupler. The phase difference determines the power ratio of the two output ports of the coupler, so that by reasonable control , the coupler can realize the output of optical power in any proportion.

[0026] The beneficial effects of the present application are:

[0027] In view of the deficiencies existing in the existing research of FSO cooperative transmission system, the present application proposes a flexible splitting FSO cooperative transmission method. For different atmospheric channel conditions, a deep reinforcement learning algorithm is used to decide different relay link sets and output optical power ratios, and the decision results are sent to the electric signal amplitude controller in real time to generate corresponding electric signals to drive the phase shifter. The flexible splitter based on the cascade of the phase shifter and the coupler is used to distribute the optical power to the relay link as needed. The present application uses the flexible splitter to adaptively select the relay link and the optical power distribution scheme, and can flexibly cope with the influence of different atmospheric channel conditions on the transmission performance.

[0028] The innovation of the flexible splitting FSO cooperative transmission method proposed by the present application lies in: solving the problems of optical power waste and transmission performance limitation caused by the fixed splitter in the traditional spatial light cooperative transmission system. Firstly, a relay link selection and optical power distribution algorithm based on deep reinforcement learning is designed; then the structure of the flexible splitter is designed, so that the input optical signal can be output to the relay link with different channel states according to any power ratio. The present application ingeniously combines the adaptive adjustment of the relay link and the optical power distribution scheme with the flexible splitter, improves the utilization rate of optical power, and improves the transmission performance of the FSO cooperative transmission system under dynamic atmospheric channel conditions.

[0029] Additional advantages, objects, and features of the application will be set forth in part by the description that follows, and in part will become apparent to those skilled in the art upon examination of the following specification or can be learned by practice of the application. The objects and other advantages of the application can be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to make the objects, technical solutions and advantages of the present application clearer, the preferred embodiments of the present application will be described in detail below with reference to the drawings, in which:

[0031] Figure 1 Schematic diagram of deep learning algorithm for relay link selection and optical power allocation

[0032] Figure 2 Flowchart of deep learning algorithm for relay link selection and optical power allocation

[0033] Figure 3 Structure diagram of fixed optical splitter cooperation FSO system with splitting ratio 1:8

[0034] Figure 4 Structure diagram of flexible optical splitter cooperation FSO system with splitting ratio 1:8

[0035] Figure 5 Average BER results when the atmospheric channel state is independent and identically distributed

[0036] Figure 6 Average BER results when the atmospheric channel state is independent and identically distributed

[0037] Figure 7 Channel attenuation cumulative distribution of different links under the condition that the atmospheric channel state is independent and identically distributed

[0038] Figure 8 Relationship between the number of relay links and the average BER under the condition that the atmospheric channel state is independent and identically distributed

[0039] Figure 9 Comparison of power waste of flexible optical splitting and fixed optical splitting in the cooperation transmission system with splitting ratio 1:8

[0040] Figure 10 Comparison of BER cumulative distribution of equal power random relay selection and flexible optical splitting adaptive relay selection based on deep learning algorithm DETAILED DESCRIPTION

[0041] The present application is further illustrated by the following specific examples, which provide further appreciation of the application by one of ordinary skill in the art. The examples described herein are illustrative only and should not be construed as limiting the scope or content of the application. The examples described herein can be combined or interchanged, where appropriate, without departing from the scope or spirit of the application.

[0042] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings:

[0043] The same or similar components in the drawings of the embodiments of the present application correspond to the same or similar components; in the description of the present application, it should be understood that if the terms "upper", "lower", "left", "right", "front", "back" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore the terms describing the positional relationship in the drawings are only for illustrative purposes, and cannot be understood as a limitation of the present application, for those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0044] The conventional cooperative transmission system mainly adopts a fixed optical splitter to divide the transmission power and broadcast it, and the corresponding receiving end adopts a spatial diversity method to receive. The more output ports of the fixed optical splitter, the greater the selectivity of the relay link, but it may also lead to more idle output ports and higher waste of power, which is to sacrifice optical power to increase the flexibility and stability of the system.

[0045] Therefore, the present application proposes a flexible optical splitter, which concentrates optical power on FSO relay links with better channel conditions as much as possible to reduce the invalid transmission of optical signals. The basic idea is to design a cascaded optical splitter structure based on phase shifters and couplers, which can realize flexible power distribution between different spatial optical links according to different spatial channel states. The specific implementation is to use an amplitude controller to control the driving voltage of the input phase modulator to realize the phase shift of the input phase modulator optical signal (shifted phase ). Then two optical signals with phase difference and equal power are coupled into the coupler, and the output power of the coupler is the sum of the two input powers, and the phase difference between the two input powers is the phase difference between the two input optical signals. As the input of the X-type coupler, it enables the output port to distribute optical power in different proportions.

[0046] Clearly, when applied to FSO cooperative transmission systems, the output optical power of the flexible optical splitter depends on the results of relay selection and power allocation. Due to the time-varying nature of atmospheric channel conditions, fixed relay selection and power allocation cannot consistently meet system performance requirements. Therefore, dynamic relay selection and power allocation are crucial for ensuring the performance of FSO cooperative transmission systems. This invention proposes an adaptive relay link selection and optical power allocation method for atmospheric channels. By acquiring atmospheric channel conditions online, a deep reinforcement learning algorithm is used to select relay links and calculate the optical power output to meet the varying optical power requirements of relay links under different channel conditions. The results of relay link selection and optical power allocation are sent to an amplitude controller to control the phase shift in the flexible optical splitter in real time, thereby adjusting the optical power output of different relay links. Finally, the system's BER (Breakpoint Error) and other information are fed back into the deep reinforcement learning algorithm for strategy improvement. Through continuous training and empirical strategy refinement, the optimal strategy for relay selection and power allocation under dynamic conditions is obtained.

[0047] like Figure 1 As shown, this invention proposes a flexible beam splitting method for cooperative FSO communication systems, and on this basis, achieves adaptive relay selection and optical power allocation. The flexible beam splitter designed in this invention can adaptively control the optical power of the relay link according to changes in atmospheric channel conditions, breaking through the performance limitations of fixed beam splitting in traditional cooperative FSO communication systems by improving optical power utilization. Therefore, this invention is of great significance for improving the performance of FSO communication systems.

[0048] This invention was verified in a cooperative FSO communication system with a split ratio of 1:8, where information is transmitted between the source node and the destination node via one direct link and seven relay links. The link numbers are as follows: The specific implementation process of the elastic beam splitter structure and the relay selection and power allocation based on deep reinforcement learning proposed in this invention is as follows:

[0049] First, a deep learning algorithm is used to adaptively select relay links and optical power allocation schemes. The algorithm flow is as follows: Figure 2 As shown, the specific steps are as follows:

[0050] 1. Define the first m Atmospheric attenuation of the link and atmospheric turbulence As a space optical channel t The channel state at a given moment. (Previous moment) t The channel state of -1, i.e. and , as the input parameters of the DQN algorithm and the DDPG algorithm, the value function and the policy function are fitted through the neural network to predict the current time t The system selects the action set of the relay link And the action set of the optical power allocation ;

[0051] 2. Send the action set to the elastic splitting cooperative transmission system, and feed back the reward value and the channel state at the current time. Store the channel state at the previous time , the action set and , the reward value and the channel state at the current time and other information in the experience pool, so as to sample and train the Q network from the experience pool.

[0052] 3. Calculate the target function Q value at the current time according to the reward value and the current channel state information, calculate the error between the predicted value of the Q network and the target function Q value, and perform back propagation to the neural network for optimization, so that it can better predict the Q value.

[0053] The main difference between the elastic splitting-based cooperative FSO transmission system and the traditional cooperative FSO transmission system in structure is the splitting part of the transmitting end, as shown in Figure 3 and Figure 4 . Figure 3 For the traditional cooperative transmission system, a fixed optical splitter divides the transmitting optical power equally into 8 output ports for broadcast transmission. Figure 4 For the elastic splitting-based cooperative transmission system, the following is the splitting process of the transmitting signal:

[0054] 1. The signal source is modulated into an NRZ (Non Return to Zero) electrical signal through an NRZ pulse generator, and the NRZ electrical signal and the optical signal output by the laser are used as the input signals of the MZM (Mach-Zehnder Modulator) for intensity modulation, so as to load the information onto the optical signal.

[0055] 2. According to the relay selection and power allocation scheme decided by the deep learning algorithm, calculate the optical phase difference required by each unit structure in the elastic optical splitter, and input the corresponding driving voltage for each phase shifter through the pulse generator.

[0056] 3. In each splitting unit inside the elastic optical splitter, the intensity-modulated optical signal is divided into two paths through a 1:1 optical splitter, and a phase shifter is used to form a phase shift between the two optical signals , and then an X-coupler is used to realize different proportions of optical power output.

[0057] After the above elastic splitting, the direct transmission link and the relay link will be inputted with optical signals of unequal power. Among them, the relay link uses EDFA to amplify and forward the received optical signal at the relay site. At the receiving end, the spatial diversity method is used for reception, and the specific demodulation method is as follows:

[0058] 1. The receiving end combines the optical signals from the direct transmission link and the relay link with equal gain.

[0059] 2. The optical signal is converted into an electrical signal by a photodetector, and a low-pass filter is used to filter out out-of-band noise.

[0060] 3. The received electrical signal is judged by a symbol decision device, and the initial information can be restored.

[0061] According to the parameter settings in Table 1, the traditional fixed splitting cooperative FSO transmission system and the elastic splitting cooperative FSO transmission system proposed in the present application are simulated by OptiSystem optical simulation software, and the system performance is compared and analyzed.

[0062] Table 1 OptiSystem optical simulation software parameter settings

[0063]

[0064] Figure 5 The BER results of the direct transmission link (only direct transmission link transmission, no relay link cooperative transmission), the fixed splitting cooperative transmission system, and the elastic splitting cooperative transmission system under the condition of no turbulence influence are shown, and the influence of different atmospheric losses of the relay link on the BER comparison results is analyzed. At this time, the atmospheric loss of the direct transmission link is fixed at 6 dB / km, and it is assumed that all relay links experience the same atmospheric loss. In order to analyze the performance advantage of the elastic splitter itself compared with the fixed splitter in cooperative transmission, Figure 5 the number of relays in the fixed splitting cooperative transmission system is selected according to Gaussian random distribution, and the transmitted optical power is equally distributed among different relay links. It can be found that as the atmospheric loss of the relay link increases, the BER of the system gradually increases. When the atmospheric loss of the relay link exceeds 4 dB / km, the fixed splitting cooperative transmission system cannot continue to obtain performance gain from spatial diversity, and its BER performance starts to be worse than that of the direct transmission link. In contrast, the elastic splitting cooperative transmission system can still achieve better BER performance than the direct transmission link at this time, and the performance advantage is sustained within the range of atmospheric loss less than 4.5 dB / km. Because the elastic splitting can better utilize the optical power, within the range of atmospheric loss less than 3.5 dB / km, the BER performance of the elastic splitting cooperative transmission system can be improved by 1 to 2 orders of magnitude compared with the fixed splitting cooperative transmission system.

[0065] According to the parameter settings in Table 2, the direct transmission link and the relay link loss are both 3.5 dB / km, wherein the atmospheric refractive index structure constant of the direct transmission link is , and the atmospheric turbulence intensity of the relay link is subject to independent identical distribution. Figure 6 The influence of atmospheric turbulence intensity on the BER performance of the two cooperative transmission systems is compared under the condition that the number of relays is randomly selected according to Gaussian distribution and the optical power is equally output. It can be seen from Figure 6 that when the atmospheric refractive index structure constant of the relay link is greater than , the fixed splitting cooperative transmission system cannot obtain obvious performance advantage compared with the direct transmission link system, while the elastic splitting cooperative transmission system can still control the BER to be below . Therefore, compared with the fixed splitting cooperative transmission system, the elastic splitting cooperative transmission system has stronger anti-atmospheric turbulence capability and can effectively improve the stability of system transmission.

[0066] Table 2 Parameter settings of the direct transmission link and the relay link

[0067]

[0068] In order to analyze the change of system BER performance under the joint influence of atmospheric loss and turbulence effect, scenes of different atmospheric loss and atmospheric turbulence of the relay link are set to be independent and identically distributed, and the random optical power decay cumulative distribution function of each relay link is shown in Figure 7 . Under the condition that the number of relays is selected according to Gaussian random distribution and the optical power is equally output, Figure 8 the influence of different number of relays on the average BER of the system is analyzed. It can be seen that with the decrease of the number of relay link selection, the BER of the two cooperative transmission systems increases. However, for any number of relays, the BER of the elastic splitting cooperative transmission system is significantly lower than that of the fixed splitting cooperative transmission system, and when the number of selected relays is less than 4, the BER performance advantage is more obvious. This shows that the elastic splitting cooperative system has higher energy efficiency and can obtain better system BER performance under the same transmission power. Taking the splitting ratio of 1:8 as an example, it can be seen from the comparison of power waste of the fixed splitting and the elastic splitting in Figure 9 that when the number of selected relays is less than 5, the elastic splitting is more power-saving than the fixed splitting, and when the splitting ratio is 1:8, the optical power can be saved by up to 37.78%.

[0069] Figure 10Cumulative distribution probabilities of BER performance of equal power random relay selection and adaptive relay selection based on deep learning algorithm are compared under time-varying conditions of atmospheric channel. When equal power random relay selection is used for relay link, under time-varying conditions of channel, the BER of the elastic cooperative transmission system using equal power random relay selection fluctuates greatly, and the system performance is unstable. In contrast, the adaptive relay selection based on deep learning algorithm can ensure that the system BER is lower than 10 -6 -10 -3 -10 as the BER threshold reference, compared with equal power random relay selection, the adaptive relay selection based on deep learning algorithm can provide more than 33% performance optimization.

[0070] In summary, the elastic light splitting space optical cooperative transmission method proposed in the present application can flexibly output optical power between different relay links according to differentiated channel quality through elastic light splitting, improve the utilization rate of optical power, and determine the channel adaptive relay selection and power allocation scheme through the deep reinforcement learning algorithm, thereby improving the performance of the space optical cooperative transmission system as a whole.

[0071] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit it. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. A method for spatial light cooperation transmission based on elastic light splitting, characterized in that: The method is as follows: The system transmitter uses a flexible optical splitter based on a cascaded phase shifter and coupler, which allows the input power to be distributed in any proportion among the different output ports of the optical splitter; Adaptive relay link selection and optical power allocation for atmospheric channels utilize deep reinforcement learning algorithms to dynamically select space optical relay links and allocate optical power to each link based on different atmospheric channel conditions, without requiring channel datasets, with the goal of minimizing the system bit error rate. The specific steps of the atmospheric channel adaptive relay link selection and optical power allocation include: 1) Under channel conditions where turbulence intensity and atmospheric loss are time-varying, the DQN algorithm is used to continuously learn and fit the optimal action value function, predict and select the set of space optical relay links with the highest Q value, denoted as the action. ; 2) Based on the time-varying state of the atmospheric channel, the Actor network simulation policy function in the DDPG algorithm is used to predict the power allocation scheme of the selected space optical relay link, denoted as the action. Using the Critic network to simulate the action value function to calculate the action of the relay link power allocation scheme. The Q-value is used to determine the quality of the Actor network's predicted actions. The goal of the policy function is to determine the action that maximizes the Q-value. 3) Relay link selection action and power distribution action Joint normalization is performed so that optical power is allocated to selected relay links in different proportions, while the optical power of unselected relay links is 0. Define the Q-value objective function for space optical relay link selection and power allocation: in for The reward value at any moment, For discount rate, A value function model fitted to a neural network. for Q value at time t, for Channel state at any given time. for The set of actions under the current channel state. Representing network weights; using an atmospheric channel as the environment for deep reinforcement learning, defining the reward function: in BER For bit error rate, when BER Less than 10 -3 The reward is positive, and BER The smaller the value, the higher the reward; after sending the relay link selection and optical power allocation scheme to the environment, the reward value and channel status are fed back in real time to update the objective function. Define the loss function: in for Q value at time t, for Channel state at any given time. for Actions under the channel conditions at any time; continuously optimize the neural network using gradient descent to minimize the loss function and maximize the Q value until the reward value converges, i.e., the cumulative expected reward is maximized; adjust the relay link and power allocation according to the channel conditions. The atmospheric channel adaptive space optical relay link selection, optical power allocation, and cooperative transmission method combining flexible optical splitters specifically include: under different FSO channel states, after selecting the relay link and output optical power using a deep reinforcement learning algorithm, the relevant information is sent to an electrical signal amplitude controller to generate a corresponding electrical signal to drive a phase shifter; the phase shift of the optical signal is achieved using a phase modulator as follows: In the above formula To output the electric field of the optical signal, For the input optical signal electric field, For phase shift, To modulate electrical signals; set up By controlling electrical signals in real time The voltage value allows for arbitrary phase shift of the input optical signal. Then, the two paths with a phase difference And optical signals of equal power are input to the coupler; according to the coupler's transfer function: （5） In the above formula and The input optical power of the coupler is denoted as , and , and The output optical power of the coupler; phase difference Determine the power ratio of the two output ports of the coupler, and control it appropriately. Achieve arbitrary ratio of optical power output from the coupler.

2. The spatial optical cooperative transmission method based on elastic beam splitting according to claim 1, characterized in that: The specific design steps for the flexible beam splitter include: 1) Encode the relay link selection and optical power allocation scheme and send it to an electrical signal amplitude controller to generate the corresponding electrical signal to drive the phase shifter; 2) The input optical signal is split into two signals using a 1:1 beam splitter. One of the optical signals is input to a phase shifter. Under the drive of the electrical signal input in step 1, the phase shift of the optical signal is achieved. 3) Input two optical signals with a phase difference into the X-coupler, and adjust the phase difference to achieve different ratios of optical power output at the two output ports; 4) Arbitrarily modify the unit structures obtained in steps 1) to 3). N Class alliance, resulting in a 1:2 ratio. N A flexible beam splitter with a specific structure achieves 2D beam splitting by inputting different driving voltages to the phase shifter array. N The optical power output of any proportion of each output port.

Images