A semiconductor laser array pulse modulation timing control method and system

Abstract

The application relates to the technical field of electric data processing, and discloses a semiconductor laser array pulse modulation time sequence control method and system, wherein the method comprises the following steps: performing adaptive exponential weighted moving average filtering on an original channel gain to obtain a smoothing vector; performing space domain regularization recursive least square prediction on the smoothing vector to obtain a predicted gain; performing maximum minimum signal-to-noise ratio time slot allocation on the predicted gain to obtain an optimal time slot index; performing digital time conversion triggering on the optimal time slot index to obtain a high-speed switch trigger signal; and performing Kalman filtering time delay compensation on the trigger signal to obtain a transmission time sequence instruction. Compared with the prior art, the method can realize real-time prediction and time sequence adaptation of channel state information, reduce the bit error rate, and improve the link reliability by cascading processing of channel filtering, prediction, allocation, triggering and compensation, and can solve the problem that the fixed modulation time sequence in the prior art cannot match channel changes, resulting in an increased bit error rate, and especially cannot realize microsecond-level pulse time sequence reconfiguration when atmospheric turbulence causes rapid fading.

Description

Technical Field

[0001] This invention relates to the field of electrical data processing technology, and in particular to a method and system for timing control of pulse modulation of a semiconductor laser array. Background Technology

[0002] Currently, in free-space laser communication systems, semiconductor laser arrays combined with pulse position modulation are commonly used to combat signal fading caused by atmospheric turbulence. Existing technologies generally allocate transmission times to each laser unit statically based on a preset fixed time slot allocation table or threshold comparisons in a slow feedback loop. The acquisition and updating of channel state information relies on traditional channel estimation methods. Subsequent electrical data processing units are mainly used for symbol encoding / decoding and modulation format mapping, lacking the closed-loop processing capability for deep filtering, trend prediction, and dynamic timing reconfiguration of real-time channel estimation data. This results in the modulation timing being unable to match the rapidly changing channel state.

[0003] For example, in a satellite-to-ground laser link, refractive index fluctuations caused by atmospheric turbulence can lead to deep fading of the receiver's light intensity on a microsecond timescale. In this case, if the fixed modulation timing still allocates time slots based on outdated channel conditions, some laser units may emit light pulses precisely at the channel gain trough, causing a sharp deterioration in the instantaneous signal-to-noise ratio at the receiver, which in turn triggers burst errors and significantly increases the bit error rate. Simultaneously, traditional methods lack efficient electrical data processing techniques, making it impossible to complete channel prediction and real-time calculation and updating of pulse timing within the coherence time. They also lack coordinated optimization of physical constraints such as array multi-unit timing and thermal crosstalk, making it difficult to guarantee the link's reliability under turbulent conditions.

[0004] Therefore, there is an urgent need for a method that can still use high-speed electrical data processing technology to filter and predict channel state information in real time, and adaptively adjust the pulse emission timing of semiconductor laser arrays, even in complex atmospheric environments with rapid fluctuations in channel gain and severe fluctuations in turbulence intensity, so as to achieve microsecond-level pulse timing reconfiguration, thereby reducing the communication bit error rate and improving the reliability and anti-turbulence capability of free space laser communication links. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the present invention aims to propose a semiconductor laser array pulse modulation timing control method. This method addresses the technical problems in existing technologies, such as the inability of fixed modulation timing to match channel changes, the lack of efficient electrical data processing methods for real-time timing updates, leading to increased bit error rates, and the inability to achieve microsecond-level pulse timing reconfiguration, especially when atmospheric turbulence causes rapid fading.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for timing control of pulse modulation of semiconductor laser array.

[0007] The semiconductor laser array pulse modulation timing control method includes: Step S10: Obtain the original channel gain vector, and perform channel preprocessing based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smooth channel vector; Step S20: Based on the smoothed channel vector, a spatial regularized recursive least squares prediction mechanism is used to perform the channel prediction task and output the predicted channel gain; Step S30: Based on the predicted channel gain, the time slot allocation task is performed using the maximum and minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and the optimal transmission time slot index is output; Step S40: Based on the optimal transmit time slot index, a digital time conversion triggering mechanism is used to perform a pulse modulation timing reconfiguration task, and a high-speed switch trigger signal is output; Step S50: Based on the high-speed switch trigger signal, a Kalman filter delay compensation mechanism is used to perform a delay feedforward compensation task, and a delay-compensated laser pulse emission timing command is output.

[0008] Preferably, step S10, which involves obtaining the original channel gain vector and performing channel preprocessing based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smoothed channel vector, specifically includes: Step S101: At the free-space laser communication receiver, the original channel gain vector corresponding to the pilot sequence is extracted through coherent detection and analog-to-digital conversion.

[0009] in At the current sampling time, This represents the total number of independently tunable laser units in a semiconductor laser array. For the first Each laser unit at the sampling time The corresponding original channel gain, where T is the vector transpose operator; Step S102: The electrical data processing unit utilizes the continuously updated background noise power during the no-signal period. Calculate the instantaneous estimated signal-to-noise ratio

[0010] in The energy of the smoothed channel vector at the previous sampling time. The square operation of the Euclidean norm 2 of a vector represents the total energy of the vector. This refers to the previous sampling time. Step S103: Use the instantaneously estimated signal-to-noise ratio to drive the adaptive smoothing factor. Dynamic updates, specifically

[0011] in , , which are the upper and lower limits of the smoothing factor. To preset the signal-to-noise ratio threshold. It is an exponential function with the natural constant e as its base; Step S104: Calculate the smoothed channel vector at the current time.

[0012] in for The original channel gain vector at time t. This is the smoothed channel gain vector from the previous time step, achieving an adaptive trade-off by increasing the smoothing factor to quickly track channel changes when the signal-to-noise ratio is high, and decreasing the smoothing factor to strongly suppress noise when the signal-to-noise ratio is low.

[0013] Preferably, step S20, which involves performing a channel prediction task based on the smoothed channel vector using a spatially regularized recursive least squares prediction mechanism and outputting the predicted channel gain, specifically includes: Step S201: For the first Each laser unit utilizes the smoothed channel vector to construct a length of... Tap delay line regression vector :

[0014] in The preset tap delay line length represents the dimension of the regression vector. , and Let T be the smoothed channel gain of the m-th laser element at sampling times k, k-1, and k-d+1, and let T be the vector transpose operator. Step S202: Update the prediction coefficient vector using recursive least squares. When using this method, the loss function is:

[0015] in The forgetting factor is adaptively determined in step S203, where i is the time-dimensional summation index variable, corresponding to all sampling times from the initial time to the k-th time, and j is the cell-dimensional summation index variable, corresponding to the set Each adjacent laser unit in the middle, Sum the terms from 0 to k for i. Forgetting factor The ki power represents historical data that is further away from the current time, with lower weights. |This is a complex modulo operation, taking the magnitude of the difference between the complex channel gain and the predicted value.| | 2 The square of the modulus corresponds to the power error. For the prediction coefficient vector w m Transpose of; In order to be with the first A set of physically adjacent laser units. For set The number of laser units in the system; This is the spatial coupling strength coefficient, which is used to constrain the consistency of the predicted coefficients by utilizing the spatial coherence of atmospheric turbulence. The input data vector; This refers to the time point corresponding to the i-th sampling moment; Let J be the prediction coefficient vector corresponding to the j-th adjacent laser unit. Step S203: Forgetting Factor By real-time flicker index Driver, specifically

[0016] in Preset configurable baseline forgetting factor parameters, It is the ratio of the variance of the received optical power within the most recent window to the square of the mean. Step S204: Calculate the future number using the updated prediction coefficient vector. Predicted channel gain of the step

[0017] Where P is the preset maximum number of prediction steps, i.e., the upper limit of the predicted future time range. for The transpose of the time prediction coefficient vector allows the forgetting factor to be automatically reduced when the scintillation index increases, i.e., when turbulence intensifies, in order to quickly forget old data, and the forgetting factor to approach 1 when turbulence is calm, so as to make full use of long-term correlation.

[0018] Preferably, in step S202, the spatial coupling strength coefficient The value is determined by the average aperture effect of the receiver and the spatial spacing of the laser array, and its value is between 0.1 and 0.5.

[0019] Preferably, step S30, which involves performing a time slot allocation task based on the predicted channel gain using a maximum-minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and outputting the optimal transmit time slot index, specifically includes: Step S301: Set the symbol period Divided into Each time slot has a width of [number] timeslots. ;No. Time slot index assigned to each laser unit When calculating its equivalent signal-to-noise ratio.

[0020] in For detector responsivity, For pulse peak power, For noise power spectral density, For receiving bandwidth, This is the starting time of the current symbol. For the first The time point corresponding to each time slot Let m be the predicted channel gain of the m-th laser unit at the target time slot; Step S302: Solve the allocation optimization problem to maximize the worst equivalent signal-to-noise ratio.

[0021] Thermal crosstalk sensing constraints are introduced during the allocation process. ,in Based on the most recent average driving current of the two laser units and A defined dynamic time slot index interval threshold, The time slot index number assigned to the a-th laser unit. The time slot index number assigned to the b-th laser unit. The absolute value of the difference between the time slot numbers of the two laser units represents the number of time slots between the emitted pulses of the two units. The smaller the value, the closer the emission times of the two pulses are, and the more significant the thermal crosstalk superposition effect. argmax is the optimization operator, which indicates the value of the independent variable that maximizes the subsequent objective function. To find the minimum value for all m-th laser units, i.e., to select the worst equivalent signal-to-noise ratio among all links, the superscript... The identifier variable is the optimal solution. The m-th laser unit is assigned to the n-th laser unit. m The equivalent received signal-to-noise ratio at the receiving end for each time slot; Step S303: When the worst equivalent signal-to-noise ratio is obtained When the prediction error covariance of the recursive least squares algorithm output is below the reliable demodulation threshold and the prediction uncertainty is less than the set threshold, the probabilistic cooperative dual-emission mechanism is activated. The laser unit with the second worst signal-to-noise ratio is selected from the remaining laser units to share the same optimal time slot with the laser unit with the worst signal-to-noise ratio, and they are emitted using orthogonal polarization states to obtain the final optimal emission time slot index.

[0022] Preferably, step S40, which involves executing a pulse modulation timing reconfiguration task based on the optimal transmit time slot index using a digital time conversion triggering mechanism and outputting a high-speed switch trigger signal, specifically includes: Step S401: The dedicated state machine verifies whether the optimal transmission time slot indices meet the thermal crosstalk sensing constraint. If the constraint is violated due to prediction deviation, it returns to the preset safe time slot table. Step S402: After the verification is passed, calculate the center time of the time slot for each laser unit. And it is quantized and written to the electrical data processing unit register in a fixed-point format, wherein The center time of the time slot corresponding to the m-th laser unit is the time offset based on the start time of the current symbol period, i.e., the center time point of the emitted pulse of this unit. This refers to the optimal time slot index number assigned to the m-th laser unit after time slot allocation optimization. The superscript indicates the index number. This value represents the optimal solution to the worst-case signal-to-noise ratio optimization problem. The time slot indices are consecutively numbered from 1 to N, corresponding to N time-division time slots within one symbol period. The time width of a single time slot; Step S403: The digital time converter generates an edge trigger signal with picosecond precision based on the written time value, directly driving the GaN high-speed switches of each laser driver stage, and outputs the high-speed switch trigger signal.

[0023] Preferably, step S50, which involves performing a delay feedforward compensation task based on the high-speed switch trigger signal using a Kalman filter delay compensation mechanism and outputting a delay-compensated laser pulse emission timing command, specifically includes: Step S501: Measure the system delay between the command issuance time of the high-speed switch trigger signal and the physical emission time of the laser pulse using a delay line phase detector, and use the system delay as an observation. Step S502: Establish a slowly varying model with the system delay as the state variable, and use a Kalman filter to track the slowly varying trend of the system delay in real time to obtain an estimate of the system delay; Step S503: Feed the estimated value of the system delay to the predictor in step S20, correct the time deviation between the predicted target time and the physical realization time, and output the laser pulse emission timing command after time delay compensation.

[0024] The present invention also provides a semiconductor laser array pulse modulation timing control system, comprising: The channel preprocessing module is used to obtain the original channel gain vector, and to perform channel preprocessing tasks based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smooth channel vector. The channel prediction module is used to perform channel prediction tasks based on the smoothed channel vector using a spatial regularized recursive least squares prediction mechanism, and output the predicted channel gain. The time slot allocation module is used to perform time slot allocation tasks based on the predicted channel gain, using the maximum and minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and output the optimal transmission time slot index. The trigger signal generation module is used to perform a pulse modulation timing reconfiguration task based on the optimal transmission time slot index using a digital time conversion triggering mechanism, and output a high-speed switching trigger signal. The delay compensation module is used to perform delay feedforward compensation tasks based on the high-speed switch trigger signal using a Kalman filter delay compensation mechanism, and output the delay-compensated laser pulse emission timing command.

[0025] The present invention also provides a semiconductor laser array pulse modulation timing control device, the semiconductor laser array pulse modulation timing control device comprising: a memory, a processor, and a semiconductor laser array pulse modulation timing control program stored in the memory and executable on the processor, wherein the semiconductor laser array pulse modulation timing control program implements the above method when executed by the processor.

[0026] The present invention also provides a computer program product, the computer program product including a semiconductor laser array pulse modulation timing control program, which implements the above method when executed by a processor.

[0027] The beneficial effects of this invention are as follows: This invention employs an adaptive exponentially weighted moving average filtering mechanism to dynamically adjust the smoothing factor based on the real-time estimated signal-to-noise ratio (SNR). This mechanism strongly suppresses noise at low SNR and rapidly tracks channel changes at high SNR, outputting a smooth channel vector that provides a stable input for subsequent predictions. This avoids the lag or noise amplification problems associated with fixed-coefficient filtering in dynamic turbulent environments.

[0028] This invention utilizes a spatial regularized recursive least squares prediction mechanism to constrain the consistency of prediction coefficients of adjacent laser units by taking advantage of the spatial coherence of atmospheric turbulence. Furthermore, it introduces a real-time scintillation index in the forgetting factor control, enabling the predictor to automatically reduce its memory length and quickly adapt to changes when turbulence intensifies. This improves the accuracy and robustness of channel prediction, thereby allowing for more reliable selection of the optimal transmission time in time slot allocation. Attached Figure Description

[0029] Figure 1This is a flowchart illustrating the first embodiment of a semiconductor laser array pulse modulation timing control method according to the present invention. Detailed Implementation

[0030] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0031] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Example 1: As Figure 1 The diagram shown is a flowchart of a first embodiment of a semiconductor laser array pulse modulation timing control method according to the present invention, which presents a first embodiment of the semiconductor laser array pulse modulation timing control method according to the present invention.

[0033] In a first embodiment, the semiconductor laser array pulse modulation timing control method includes: Step S10: Obtain the original channel gain vector, and perform channel preprocessing based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smooth channel vector; This step receives the original channel gain vector corresponding to the pilot sequence extracted by coherent detection and analog-to-digital conversion at the receiver. ,in This represents the total number of independently tunable laser units in a semiconductor laser array. This is the current sampling time. The electrical data processing unit first utilizes the continuously updated background noise power during the no-signal period. Combined with the energy of the smoothed channel vector at the previous sampling time Calculate the instantaneous estimated signal-to-noise ratio Then, the adaptive smoothing factor is driven by this instantaneously estimated signal-to-noise ratio. Dynamic updates: ,in , , A preset signal-to-noise ratio threshold is set. Finally, an exponentially weighted recursive method is used. The output is a smoothed channel vector. The core of this adaptive filtering is to adjust the value of the smoothing factor in real time according to the channel quality, thereby providing the subsequent spatial regularized recursive least squares prediction with an input time series that is sufficiently noise-suppressed and can reflect channel changes in a timely manner.

[0034] After adaptive exponentially weighted moving average filtering, the high-frequency jitter caused by detector thermal noise and atmospheric scintillation in the original channel gain is compressed, while the effective components reflecting the changing trends of large-scale turbulent fading and cloud obstruction are preserved. In the deep fading region, the instantaneous estimated signal-to-noise ratio decreases, and the smoothing factor automatically approximates... The filter increases the weight of historical smoothing values ​​to prevent noise from being misinterpreted as channel changes, thus preventing the introduction of spurious high-frequency components into the subsequent predictor's regression vector. When channel quality improves, the smoothing factor increases to approximately... This assigns greater weight to the current measurement value, enabling the filtered output to quickly follow the rapid fluctuations in gain. This signal-to-noise ratio-driven adaptive smoothing ensures that the smoothed channel vector output to step S20 provides a stable foundation for time-series modeling without losing the latest trend information needed for prediction due to over-smoothing, directly improving the convergence stability of the condition number of the data matrix and parameter estimation in the recursive least squares algorithm.

[0035] Traditional channel preprocessing often uses moving averages with fixed coefficients or infinite impulse response filters with constant forgetting factors. In the turbulent atmospheric environment of free-space laser communication, a fixed filtering strength is insufficient to balance noise suppression and tracking speed: a large smoothing coefficient results in significant noise residue at low signal-to-noise ratios, leading to increased variance in subsequent recursive least-squares predictions; a small smoothing coefficient causes filter response lag as turbulence intensifies, causing a sharp increase in prediction error. This step utilizes instantaneous signal-to-noise ratio estimation to adjust the smoothing factor online, eliminating the need for manual calibration and automatically matching appropriate filtering characteristics to dynamically changing channel conditions. Within the same atmospheric environmental variation range, the smoothed channel vector exhibits better mean square error and phase lag than the fixed-coefficient method, providing a more robust data front-end for the entire timing control scheme, thereby contributing to improved real-time performance and accuracy of pulse timing reconfiguration.

[0036] For example, in a free-space laser communication, the original channel gain vector extracted by the receiver exhibits high-frequency jitter due to local wind shear turbulence superimposed with detector thermal noise. The background noise power updated during the signal-free period is a known value, and calculations show that the instantaneous estimated signal-to-noise ratio drops to... If the value is below the preset threshold, the smoothing factor automatically decreases to approximately [value missing]. The filtered output shows a significant reduction in smooth channel vector jitter, retaining only the approximate low-frequency fading envelope. Subsequently, turbulence decreases, and the signal-to-noise ratio recovers. The smoothing factor increased rapidly to Nearby, the filter output closely follows the short-term rise and fall of the channel gain, providing elements that are both smooth and synchronized with the current channel evolution for the next step of the predictor to construct the tap delay line regression vector, so that each tap value in the regression vector can truly reflect the channel pattern over the recent period.

[0037] Step S20: Based on the smoothed channel vector, a spatial regularized recursive least squares prediction mechanism is used to perform the channel prediction task and output the predicted channel gain; This step takes the smoothed channel vector as input and targets the... Each laser unit, first utilizing its current and past... The length of the smoothed channel gain value constructed at each time step is... Tap delay line regression vector Prediction coefficient vector By using recursive least squares updates, its loss function adds a spatial regularization penalty term to the standard recursive least squares: .in Is with the first A set of physically adjacent laser units. This is the spatial coupling strength coefficient, and its value typically ranges from [value range missing]. to Forgetting factor It is no longer fixed, but determined by the real-time flicker index. drive: ,in The updated coefficients are used to calculate the future... Predicting Channel Gain It is directly used for time slot allocation in step S30.

[0038] After introducing spatial regularization, the update of recursive least squares coefficients depends not only on the historical patterns of the current cell's channel but also on the channel behavior of spatially adjacent cells. Within the spatial coherence length, atmospheric turbulence causes adjacent laser cells to experience similar channel variations. The spatial penalty term effectively suppresses abnormal coefficient deviations caused by single-channel noise or transient fluctuations by guiding the current cell's coefficients towards the mean of adjacent cell coefficients, thus enhancing the spatial smoothness and numerical stability of the prediction results. Simultaneously, the forgetting factor is adaptively adjusted by the scintillation exponent: when turbulence intensifies and the scintillation exponent increases, As the baseline value decreases, the algorithm accelerates the forgetting of outdated data to track non-stationary changes in the channel; when the turbulence is smooth... Approaching This approach fully leverages long-term correlation to improve prediction accuracy. The resulting predicted channel gain achieves a better balance between the rate of change and prediction bias, providing a more accurate channel state expectation for time slot allocation in step S30 to maximize the worst equivalent signal-to-noise ratio.

[0039] Traditional independent recursive least squares channel prediction processes each laser unit individually, ignoring the spatial correlation between array units. In the presence of spatial coherence in atmospheric turbulence, the prediction coefficients of each unit may diverge out of order, especially when the signal-to-noise ratio of a particular unit temporarily deteriorates, causing significant oscillations in the prediction results and misleading time slot allocation. This step incorporates spatial information into the coefficient update process through regularization, ensuring physically reasonable spatial consistency of the prediction array without significantly increasing pilot overhead or computational complexity. This reduces the probability of further deterioration of worst-case link performance due to single-point prediction errors. Furthermore, traditional recursive least squares with a fixed forgetting factor blindly relies on memory length when turbulence intensity changes, while this step's adaptive forgetting factor improves prediction robustness across multiple scenarios, helping to address signal fading and bit error rate issues caused by atmospheric turbulence.

[0040] For example, in an array containing four laser units, the second unit is affected by local turbulence, causing the scintillation index to suddenly rise to [a certain value]. The predictor in this unit automatically detects the forgetting factor from approximately... Down to The system begins to rapidly forget earlier, stable-period data, and the prediction coefficients are quickly adjusted to match the sharp fluctuations of recent moments. Simultaneously, the adjacent first and third units also experience similar fading. Spatial regularization forces the prediction coefficients of the second unit to be pulled towards the mean of the coefficients of the first and third units, avoiding overfitting and extreme jumps due to excessively short data windows. As a result, the deviation between the predicted channel gain of the second unit for the future third step and the actual value is controlled within a certain range. Within this range, it is much smaller than when no spatial regularization is applied. The deviation provides a reliable reference for subsequent time slot allocation.

[0041] Step S30: Based on the predicted channel gain, the time slot allocation task is performed using the maximum and minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and the optimal transmission time slot index is output; This step allocates time slots based on the predicted channel gain of each laser unit in future time slots. Let the symbol period be... Divided into Each time slot has a width of [number] timeslots. . No. If a laser unit is assigned a time slot index Its equivalent signal-to-noise ratio is calculated as follows: ,in For detector responsivity, For pulse peak power, For noise power spectral density, For receiving bandwidth, This represents the initial time of the current symbol. Next, we solve the allocation optimization problem that maximizes the worst equivalent signal-to-noise ratio. At the same time, thermal crosstalk sensing constraints are introduced. ,in It is based on the most recent average driving current of the two laser units. and A defined dynamic time slot index interval threshold. When the worst equivalent signal-to-noise ratio is obtained. When the error covariance of the recursive least squares predictor indicates low prediction uncertainty, the probabilistic cooperative dual-transmission mechanism is activated. The second-worst signal-to-noise ratio unit and the worst unit share the same optimal time slot and are transmitted using orthogonal polarization state to obtain the final optimal transmission time slot index, which is then passed to step S40 to generate a trigger signal.

[0042] After employing the maximum and minimum signal-to-noise ratio (SNR) allocation, the system performance bottleneck is concentrated on the worst-performing link. By strategically allocating the time slot with the best channel conditions to the laser unit with the lowest current SNR, the reception quality of the worst-performing link can be improved, thereby directly reducing the overall bit error rate of the system. The introduction of thermal crosstalk constraints avoids the dense operation of multiple units in adjacent or near-adjacent time slots due to simply pursuing SNR, alleviating wavelength drift and power drop caused by local heat accumulation and ensuring the stability of the laser's operating point. The cooperative dual-transmission mechanism, when a single link is still insufficient to meet demodulation requirements, utilizes another relatively better link to transmit simultaneously with orthogonal polarization, providing diversity gain without additional time slot overhead. This allows the receiver to obtain spatially redundant copies for merging, mitigating the impact of deep fading. The entire allocation result directly provides an optimized and thermally safe optimal transmission time slot index for step S40.

[0043] Traditional pulse slot allocation often employs polling or fixed mapping based on average signal-to-noise ratio, failing to provide differentiated services for real-time channel changes. This results in wasted potential for cells with good channel conditions, while cells with poor channel conditions experience frequent bit errors. In semiconductor laser arrays, ignoring thermal crosstalk constraints and implementing high-density timing arrangements can lead to localized temperature increases in the chip, causing wavelength drift and power drops, ultimately degrading communication quality. This approach models slot allocation as a constrained optimization problem and incorporates dynamic thermal sensing and prediction reliability assessment into the solution process. This not only achieves real-time optimal matching of channel resources but also ensures hardware operational safety. Compared to traditional methods, this approach improves the problem of improper timing configuration caused by rapid changes in atmospheric turbulence channels, providing more reliable microsecond-level pulse timing reconfiguration capabilities for free-space laser communication links.

[0044] For example, suppose the symbol period is divided into equal parts. One time slot, four laser units in one unit The predicted channel gain is extremely low, only [amount missing]. After optimization, the element... Assigned to prediction gain up to The In the time slot, its equivalent signal-to-noise ratio is improved to above the demodulation threshold. Simultaneously, the unit... and Due to their proximity and high average drive current, the dynamic thermal crosstalk threshold is set to... For each time slot, the solver ensures that the difference between their time slot indices is at least 1. This avoids thermal crosstalk. In another instance of strong turbulence, the unit... Even after optimization, the worst equivalent signal-to-noise ratio remains below the threshold. However, the covariance of the recursive least squares prediction error is small, indicating reliable prediction. Therefore, the system activates cooperative dual-transmission and selects the unit with the second worst signal-to-noise ratio. Also in the The time slots transmit the same data with orthogonal polarization. The receiver combines the two signals through polarization diversity, resulting in a lower bit error rate compared to a single-transmission unit. It drops by about an order of magnitude.

[0045] Step S40: Based on the optimal transmit time slot index, a digital time conversion triggering mechanism is used to perform a pulse modulation timing reconfiguration task, and a high-speed switch trigger signal is output; After receiving the optimal transmit time slot index output from step S30, this step first uses a dedicated state machine to verify whether the thermal crosstalk sensing constraints are satisfied between the indices. Although the allocation in step S30 already considers constraints, minor conflicts may still occur during discrete time slot index quantization due to prediction bias or discretization effects. Once a constraint violation is detected, the state machine immediately reverts to the pre-stored safe time slot table to ensure hardware safety. After successful verification, the time slot index is allocated to each laser unit. Calculate its time slot center time The time value is quantized and written into the corresponding timing register in the electrical data processing unit in a fixed-point format. The digital time converter generates an edge trigger signal with picosecond precision based on the written time value. This signal directly drives the GaN high-speed switches of each laser driver stage, and finally outputs a high-speed switch trigger signal, which is provided to step S50 for time delay compensation and final pulse emission.

[0046] By employing a hard real-time triggering link combining a dedicated state machine and a digital time converter, the entire process from slot index parsing to physical trigger signal generation is completed at the hardware level. This eliminates the need for microprocessor interrupts or non-deterministic scheduling by the operating system, ensuring the accuracy and periodic consistency of each pulse transmission. Picosecond-level time resolution minimizes pulse position deviation within a specified slot, helping to reduce inter-symbol interference and inter-slot energy leakage caused by trigger jitter. The secondary verification of thermal crosstalk constraints and the safety table backoff mechanism provide hardware-level protection. Even if the upper-layer slot allocation algorithm provides an overly dense slot configuration due to occasional anomalies, it can promptly switch to a safe timing scheme, protecting the laser from thermal overload. This provides a low-jitter and safe trigger signal reference for the subsequent delay compensation step S50.

[0047] Traditional pulse timing generation often relies on the general-purpose input / output pins or general-purpose timer modules of embedded processors. Their trigger signals are limited by interrupt response delays, instruction execution cycles, and clock precision, resulting in random jitter of tens of nanoseconds or even larger. At gigabit rates, this can lead to slot positioning errors and increase the bit error rate. Furthermore, traditional methods lack online hardware verification against thermal crosstalk. When the time slots provided by the algorithm are too dense, the lack of hardware interception can easily cause the laser junction temperature to rise, threatening device lifespan. This step, through a hard real-time triggering link composed of a dedicated state machine and a digital time converter, transfers the main decision-making power of timing control from software to hardware, achieving repeatable picosecond-level precision triggering. It also incorporates a thermal safety strategy to reduce trigger uncertainty and improve the overall accuracy of timing control.

[0048] For example, the time slot allocation module provides the optimal time slot indices for the four units as follows: , , , During state machine verification, the unit was found to be... (index ) and unit (index If the time slot center time difference is insufficient to meet the current dynamic threshold for thermal crosstalk, the conflict cannot be resolved through fine-tuning. The state machine immediately discards the current allocation result and instead calls the pre-stored safe time slot table. , , , This ensures the overall heat load is distributed. The digital time converter generates a trigger signal based on the center time of the corresponding time slot of the safety meter. Under normal operating conditions where the safety meter is not activated, the fixed-point quantized time value is accurately written into the register, and the deviation between the trigger edge generated by the digital time converter and the ideal time slot center does not exceed [a certain value]. Picosecond high-speed GaN switches complete conduction in an extremely short time, injecting pulse current with regular waveform and accurate timing, providing a stable timing reference for subsequent delay compensation.

[0049] Step S50: Based on the high-speed switch trigger signal, a Kalman filter delay compensation mechanism is used to perform a delay feedforward compensation task, and a delay-compensated laser pulse emission timing command is output.

[0050] This step focuses on the system delay between the command given by the high-speed switch trigger signal and the physical emission of the laser pulse. This delay includes the delay of the drive circuit group, the amplifier response time, and the laser's spontaneous emission settling time. The delay line phase detector measures the system delay in real time as an observation by comparing the logic edge of the trigger signal with the edge of the actual emitted pulse monitored by the photodetector. The Kalman filter uses this system delay as a state variable to establish a state-space model (such as a random walk or a first-order Markov model) describing its slowly varying characteristics, and uses continuous observations to recursively estimate the smoothed value of the system delay in real time. This estimate is fed forward to the channel predictor in step S20 to correct the time deviation between the predicted target time and the physical realization time, i.e., to advance or delay the reference of the predicted time. Finally, it outputs a time-delay compensated laser pulse emission timing command to control each laser unit to emit pulses at the accurate physical time.

[0051] After introducing Kalman filter delay compensation, even if the system delay slowly drifts due to factors such as temperature changes and device aging, the future moment targeted by the predictor will always correspond to the actual physical transmission moment, eliminating the cumulative impact of delay drift on time slot alignment accuracy. The dynamic delay estimation provided by the Kalman filter has the ability to suppress single measurement noise and track slowly changing trends. Compared with directly using the original observations for compensation, its output is smoother and more stable, avoiding additional timing jitter caused by overcompensation. By feeding the delay estimate forward to the channel predictor in step S20, the entire timing control loop forms a feedforward-feedback composite structure from command generation to physical execution, ensuring that the laser pulse can accurately fall within the allocated time slot window and guaranteeing strict synchronization between the receiver sampling time and the transmission pulse.

[0052] In traditional methods, system delay is typically compensated once by a fixed value calibrated at the factory, or manually updated after open-loop measurement when the system is shut down. This approach cannot handle the gradual changes in delay caused by temperature variations and device parameter drift during operation. Compensation deviations accumulate over time, eventually causing pulses to shift within time slots or even fall into adjacent time slots, leading to inter-symbol crosstalk and increased bit error rate. This step utilizes a Kalman filter for online closed-loop estimation of delay, maintaining the accuracy of delay compensation without interrupting normal communication. This is particularly suitable for the stable operation of unattended free-space laser communication links for extended periods. By feeding delay information forward to the prediction stage, the uncertainty of physical delay is further incorporated into timing planning, overcoming the error accumulation defects of traditional fixed compensation schemes in dynamic environments.

[0053] For example, on an outdoor free-space laser communication link, the system delay of the drive circuit and laser is affected by the temperature difference between day and night, from the initial... Nanoseconds slowly drift to Nanoseconds. The delay line phase detector outputs a measured delay value once per symbol period. The Kalman filter, based on preset slowly varying model parameters, suppresses sub-nanosecond noise in a single measurement, yielding a smooth estimate of the current delay. Nanoseconds. This estimate is fed back to the channel predictor in real time. When calculating the channel gain at future times, the predictor advances the original time index by approximately [number missing]. The nanosecond time ensures that the actual generation time of the laser pulse is precisely aligned with the center of the allocated time slot after the high-speed switch trigger signal is emitted. The eye diagram opening at the receiving end is not affected by delay drift, the communication bit error rate remains stable, and the link can maintain good synchronization performance even in environments with temperature variations.

[0054] Example 2: Furthermore, the semiconductor laser array pulse modulation timing control system provided by the present invention employs a semiconductor laser array pulse modulation timing control method from the above embodiments, which can solve a technical problem in semiconductor laser array pulse modulation timing control. The beneficial effects of the semiconductor laser array pulse modulation timing control system provided by the present invention are the same as those of the semiconductor laser array pulse modulation timing control method from the above embodiments, and other technical features in the semiconductor laser array pulse modulation timing control system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0055] Example 3: This invention provides a semiconductor laser array pulse modulation timing control device. The device includes at least one processor and a memory communicatively connected to the processor. The memory stores instructions executable by the processor, which are then executed to enable the processor to perform the semiconductor laser array pulse modulation timing control method described in Example 1. This semiconductor laser array pulse modulation timing control device can include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This semiconductor laser array pulse modulation timing control device is merely an example and should not limit the functionality or scope of the invention. The device may also include a processing unit (e.g., a central processing unit, a graphics processing unit), which can perform various appropriate actions and processes based on a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of a semiconductor laser array pulse modulation timing control device. The processing unit, read-only memory, and random access memory are interconnected via a bus. The I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the semiconductor laser array pulse modulation timing control device to communicate wirelessly or wiredly with other devices to exchange data. While a semiconductor laser array pulse modulation timing control device with various systems has been described, it should be understood that it is not required to implement or possess all the systems described. Alternatively, more or fewer systems can be implemented.

[0056] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the semiconductor laser array pulse modulation timing control method described above. The computer program product provided by this invention can solve a technical problem related to semiconductor laser array pulse modulation timing control. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the semiconductor laser array pulse modulation timing control method provided in the above embodiments, and will not be repeated here.

[0057] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0058] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0059] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for timing control of pulse modulation of a semiconductor laser array, characterized in that, The methods include: Step S10: Obtain the original channel gain vector, and perform channel preprocessing based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smooth channel vector; Step S20: Based on the smoothed channel vector, a spatial regularized recursive least squares prediction mechanism is used to perform the channel prediction task and output the predicted channel gain; Step S30: Based on the predicted channel gain, a time slot allocation task is performed using a maximum and minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and the optimal transmission time slot index is output; Step S40: Based on the optimal transmit time slot index, a digital time conversion triggering mechanism is used to perform a pulse modulation timing reconfiguration task, and a high-speed switch trigger signal is output; Step S50: Based on the high-speed switch trigger signal, a Kalman filter delay compensation mechanism is used to perform a delay feedforward compensation task, and a delay-compensated laser pulse emission timing command is output.

2. The semiconductor laser array pulse modulation timing control method as described in claim 1, characterized in that, Step S10, which involves obtaining the original channel gain vector and performing channel preprocessing based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smoothed channel vector, specifically includes: Step S101: At the free-space laser communication receiver, the original channel gain vector corresponding to the pilot sequence is extracted through coherent detection and analog-to-digital conversion. for: in At the current sampling time, This represents the total number of independently tunable laser units in a semiconductor laser array. For the first Each laser unit at the sampling time The corresponding original channel gain, where T is the vector transpose operator; Step S102: The electrical data processing unit utilizes the continuously updated background noise power during the no-signal period. Calculate the instantaneous estimated signal-to-noise ratio : in The energy of the smoothed channel vector at the previous sampling time. The square operation of the Euclidean norm 2 of a vector represents the total energy of the vector. This refers to the previous sampling time. Step S103: Use the instantaneously estimated signal-to-noise ratio to drive the adaptive smoothing factor. Dynamic updates, specifically: in , , which are the upper and lower limits of the smoothing factor. To preset the signal-to-noise ratio threshold. It is an exponential function with the natural constant e as its base; Step S104: Calculate the smoothed channel vector at the current time. : in for The original channel gain vector at time t. This is the smoothed channel gain vector from the previous time step, achieving an adaptive trade-off by increasing the smoothing factor to quickly track channel changes when the signal-to-noise ratio is high, and decreasing the smoothing factor to strongly suppress noise when the signal-to-noise ratio is low.

3. The semiconductor laser array pulse modulation timing control method as described in claim 1, characterized in that, Step S20, which involves performing channel prediction based on the smoothed channel vector using a spatially regularized recursive least squares prediction mechanism and outputting the predicted channel gain, specifically includes: Step S201: For the first Each laser unit utilizes the smoothed channel vector to construct a length of... Tap delay line regression vector : in The preset tap delay line length represents the dimension of the regression vector. , and Let T be the smoothed channel gain of the m-th laser element at sampling times k, k-1, and k-d+1, and let T be the vector transpose operator. Step S202: Update the prediction coefficient vector using recursive least squares. When using this method, the loss function is: in The forgetting factor is adaptively determined in step S203, where i is the time-dimensional summation index variable, corresponding to all sampling times from the initial time to the k-th time, and j is the cell-dimensional summation index variable, corresponding to the set Each adjacent laser unit in the middle, Sum the terms from 0 to k for i. Forgetting factor The ki power represents historical data that is further away from the current time, with lower weights. |This is a complex modulo operation, taking the magnitude of the difference between the complex channel gain and the predicted value.| | 2 The square of the modulus corresponds to the power error. For the prediction coefficient vector w m Transpose of; In order to be with the first A set of physically adjacent laser units. For set The number of laser units in the system; This is the spatial coupling strength coefficient, which is used to constrain the consistency of the predicted coefficients by utilizing the spatial coherence of atmospheric turbulence. The input data vector; This refers to the time point corresponding to the i-th sampling moment; Let J be the prediction coefficient vector corresponding to the j-th adjacent laser unit. Step S203: Forgetting Factor By real-time flicker index Driver, specifically in Preset configurable baseline forgetting factor parameters, It is the ratio of the variance of the received optical power within the most recent window to the square of the mean. Step S204: Calculate the future number using the updated prediction coefficient vector. Predicted channel gain of the step Where P is the preset maximum number of prediction steps, i.e., the upper limit of the predicted future time range. for The transpose of the time prediction coefficient vector allows the forgetting factor to be automatically reduced when the scintillation index increases, i.e., when turbulence intensifies, in order to quickly forget old data, and the forgetting factor to approach 1 when turbulence is calm, so as to make full use of long-term correlation.

4. The semiconductor laser array pulse modulation timing control method as described in claim 3, characterized in that, In step S202, the spatial coupling strength coefficient The value is determined by the average aperture effect of the receiver and the spatial spacing of the laser array, and its value is between 0.1 and 0.

5.

5. The semiconductor laser array pulse modulation timing control method as described in claim 1, characterized in that, Step S30, which involves performing a time slot allocation task based on the predicted channel gain using a maximum-minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and outputting the optimal transmit time slot index, specifically includes: Step S301: Set the symbol period Divided into Each time slot has a width of [number] timeslots. ;No. Time slot index assigned to each laser unit When calculating its equivalent signal-to-noise ratio. in For detector responsivity, For pulse peak power, For noise power spectral density, For receiving bandwidth, This is the starting time of the current symbol. For the first The time point corresponding to each time slot Let m be the predicted channel gain of the m-th laser unit at the target time slot; Step S302: Solve the allocation optimization problem to maximize the worst equivalent signal-to-noise ratio. Thermal crosstalk sensing constraints are introduced during the allocation process. ,in Based on the most recent average driving current of the two laser units and A defined dynamic time slot index interval threshold, The time slot index number assigned to the a-th laser unit. The time slot index number assigned to the b-th laser unit. The absolute value of the difference between the time slot numbers of the two laser units represents the number of time slots between the emitted pulses of the two units. The smaller the value, the closer the emission times of the two pulses are, and the more significant the thermal crosstalk superposition effect. argmax is the optimization operator, which indicates the value of the independent variable that maximizes the subsequent objective function. To find the minimum value for all m-th laser units, i.e., to select the worst equivalent signal-to-noise ratio among all links, the superscript... The identifier variable is the optimal solution. The m-th laser unit is assigned to the n-th laser unit. m The equivalent received signal-to-noise ratio at the receiving end for each time slot; Step S303: When the worst equivalent signal-to-noise ratio is obtained When the prediction error covariance of the recursive least squares algorithm output is below the reliable demodulation threshold and the prediction uncertainty is less than the set threshold, the probabilistic cooperative dual-emission mechanism is activated. The laser unit with the second worst signal-to-noise ratio is selected from the remaining laser units to share the same optimal time slot with the laser unit with the worst signal-to-noise ratio, and they are emitted using orthogonal polarization states to obtain the final optimal emission time slot index.

6. The semiconductor laser array pulse modulation timing control method as described in claim 1, characterized in that, Step S40, which involves executing a pulse modulation timing reconfiguration task based on the optimal transmit time slot index using a digital time conversion triggering mechanism and outputting a high-speed switch trigger signal, specifically includes: Step S401: The dedicated state machine verifies whether the optimal transmission time slot indices meet the thermal crosstalk sensing constraint. If the constraint is violated due to prediction deviation, it returns to the preset safe time slot table. Step S402: After the verification is passed, calculate the time slot center time of each laser unit. And it is quantized and written to the electrical data processing unit register in a fixed-point format, wherein The center time of the time slot corresponding to the m-th laser unit is the time offset based on the start time of the current symbol period, i.e., the center time point of the emitted pulse of this unit. This refers to the optimal time slot index number assigned to the m-th laser unit after time slot allocation optimization. The superscript indicates the index number. This value represents the optimal solution to the worst-case signal-to-noise ratio optimization problem. The time slot indices are consecutively numbered from 1 to N, corresponding to N time-division time slots within one symbol period. The time width of a single time slot; Step S403: The digital time converter generates an edge trigger signal with picosecond precision based on the written time value, directly driving the GaN high-speed switches of each laser driver stage, and outputs the high-speed switch trigger signal.

7. The semiconductor laser array pulse modulation timing control method as described in claim 1, characterized in that, Step S50, which involves performing a delay feedforward compensation task based on the high-speed switch trigger signal using a Kalman filter delay compensation mechanism and outputting a delay-compensated laser pulse emission timing command, specifically includes: Step S501: Measure the system delay between the command issuance time of the high-speed switch trigger signal and the physical emission time of the laser pulse using a delay line phase detector, and use the system delay as an observation. Step S502: Establish a slowly varying model with the system delay as the state variable, and use a Kalman filter to track the slowly varying trend of the system delay in real time to obtain an estimate of the system delay; Step S503: Feed the estimated value of the system delay to the predictor in step S20, correct the time deviation between the predicted target time and the physical realization time, and output the laser pulse emission timing command after time delay compensation.

8. A semiconductor laser array pulse modulation timing control system, applied to the semiconductor laser array pulse modulation timing control method according to any one of claims 1 to 7, characterized in that, The system includes: The channel preprocessing module is used to obtain the original channel gain vector, and to perform channel preprocessing tasks based on the original channel gain vector using an adaptive exponential weighted moving average filtering mechanism to output a smooth channel vector. The channel prediction module is used to perform channel prediction tasks based on the smoothed channel vector using a spatial regularized recursive least squares prediction mechanism, and output the predicted channel gain. The time slot allocation module is used to perform time slot allocation tasks based on the predicted channel gain, using the maximum and minimum signal-to-noise ratio time slot allocation and cooperative enhancement mechanism, and output the optimal transmission time slot index. The trigger signal generation module is used to perform a pulse modulation timing reconfiguration task based on the optimal transmission time slot index using a digital time conversion triggering mechanism, and output a high-speed switching trigger signal. The delay compensation module is used to perform delay feedforward compensation tasks based on the high-speed switch trigger signal using a Kalman filter delay compensation mechanism, and output the delay-compensated laser pulse emission timing command.

9. A semiconductor laser array pulse modulation timing control device, characterized in that, The semiconductor laser array pulse modulation timing control device includes: a memory, a processor, and a semiconductor laser array pulse modulation timing control program stored in the memory and executable on the processor. When the semiconductor laser array pulse modulation timing control program is executed by the processor, it implements a semiconductor laser array pulse modulation timing control method according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a semiconductor laser array pulse modulation timing control program, which, when executed by a processor, implements a semiconductor laser array pulse modulation timing control method according to any one of claims 1 to 7.

Images