A signal processing method and apparatus for a quantum key distribution system

Abstract

This invention discloses a signal processing method for a quantum key distribution system, including step one, multi-dimensional signal sensing and preprocessing; step two, adaptive noise suppression; step three, phase and clock synchronization calibration; and step four, efficient key processing and post-processing. The device includes a signal sensing module, an adaptive processing module, a synchronization calibration module, and a key processing module. This invention effectively improves the signal-to-noise ratio of quantum signals and reduces the bit error rate through multi-dimensional noise sensing and adaptive Wiener filtering. It achieves high-precision phase and clock synchronization through closed-loop feedback control and uses a parallel computing architecture for computation, which effectively improves data processing efficiency and meets the real-time requirements of high-speed QKD systems. Furthermore, the invention combines machine learning prediction and closed-loop feedback control, which can adaptively respond to dynamic changes in the optical fiber channel, thus improving robustness.

Description

Technical Field

[0001] This invention relates to the field of equipment technology, and in particular to a signal processing method and apparatus for a quantum key distribution system. Background Technology

[0002] In the signal processing methods of quantum key distribution, spontaneous emission noise in optical fiber channels, nonlinear phase noise caused by the Kerr effect, and polarization drift caused by environmental vibrations can significantly reduce the signal-to-noise ratio of single-photon detection and increase the bit error rate. Meanwhile, optical fiber phase drift during long-distance transmission can lead to phase decoding errors, and traditional time synchronization methods suffer from delay fluctuations and complex deployment issues.

[0003] To address the aforementioned issues, existing technologies generally employ passive noise suppression to mitigate noise interference and improve the signal-to-noise ratio, but this method sacrifices photon utilization. Active phase compensation addresses phase drift, but it suffers from response delays and cannot adapt to rapidly changing channels. Furthermore, time synchronization relies on external devices, increasing system complexity. Therefore, this invention proposes a signal processing method and apparatus for a quantum key distribution system to solve the problems existing in the prior art. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to propose a signal processing method and apparatus for a quantum key distribution system. This method and apparatus solves the problems of low key generation rate and poor stability of existing QKD systems under long-distance, high-noise channels by employing multi-dimensional signal sensing, adaptive noise suppression, phase / time synchronization calibration, and efficient key processing, thereby significantly improving the practicality and security of the system.

[0005] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a signal processing method for a quantum key distribution system, comprising the following steps:

[0006] Step 1: Multi-dimensional signal perception and preprocessing. Noise characteristics of the sensing quantum signal channel are collected through the auxiliary detection channel to screen valid events and calibrate timestamps.

[0007] Step 2: Adaptive noise suppression. Based on the perceived noise features, a Wiener filter and machine learning model are constructed to suppress phase noise, polarization noise, and intensity noise.

[0008] Step 3: Phase and clock synchronization calibration, using closed-loop feedback control to optimize and control phase synchronization error and clock synchronization delay fluctuations;

[0009] Step 4: Efficient key processing and post-processing. An architecture using basis vector comparison, low-density parity check code and SHA-3 hash function parallel computation is used to extract the original key, correct errors and amplify privacy before outputting the secure key.

[0010] Further improvements are made in the following steps: In step one, noise characteristics of the sensing quantum signal channel are collected through an auxiliary detection channel, and quantum signal components are extracted based on wavelet transform and blind source separation algorithms. Threshold discrimination and basis vector comparison are performed on the original electrical pulse signal output by the single-photon detector to screen valid events, eliminate misjudged "vacuum pulses" and "multiphoton pulses", and use hardware phase-locked loop and bidirectional fiber time transfer technology to calibrate the timestamp error between the transmitting end and the receiving end.

[0011] The further improvement lies in the fact that the noise characteristics are perceived in real time by collecting Rayleigh scattering noise of the optical fiber channel through continuous wave monitoring light at the receiving end, using wavelet transform to perform time-frequency decomposition of quantum signal and noise, and combining blind source separation algorithm to separate quantum signal components.

[0012] A further improvement is made in that: the calibration of the timestamp error is specifically achieved by first deploying hardware phase-locked loops at the transmitting end and the receiving end to generate local synchronization clocks, then the transmitting end periodically sends optical pulses with timestamps, the receiving end records the pulse arrival time through a high-speed time-to-digital converter, and finally uses bidirectional fiber optic time transfer technology to calculate and calibrate the clock offset, and calibrates the timestamp based on the obtained offset.

[0013] The further improvement lies in the following: the phase noise suppression in step two is based on the real-time sensing phase noise power spectral density to design an adaptive Wiener filter with dynamically adjustable center frequency and bandwidth, then an LSTM neural network is introduced to predict short-term fluctuations in phase noise, and then the coefficients of the Wiener filter are adjusted according to the prediction results.

[0014] The suppression of polarization noise is achieved by splitting the quantum signal into two orthogonal polarization components using a polarization beam splitter, monitoring the intensity ratio of the two components, optimizing the voltage parameters of the programmable polarization controller using a gradient descent algorithm, and finally minimizing the polarization extinction ratio loss.

[0015] Suppressing intensity noise involves applying the Kalman filter algorithm to the electrical pulse sequence of a single-photon detector and establishing state transition equations and observation equations. Then, the event sequence is smoothed through a prediction-correction process, and an abnormal pulse judgment threshold is set to eliminate abnormal pulses.

[0016] Further improvements are made in the following steps: In step three, a phase reference driven by a temperature-controlled crystal oscillator is set at the receiving end. By comparing the deviation between the actual phase of the quantum signal and the reference phase, the output of the local phase modulator is adjusted in real time using a PID controller to optimize the phase synchronization error. Then, at the transmitting end, time-stamped optical pulses are periodically sent. At the receiving end, the delay is calculated by combining the fiber group velocity dispersion model and the local clock offset is dynamically updated to optimize the clock synchronization delay fluctuation.

[0017] The further improvement lies in the following: In step four, the extraction of the original key specifically involves both parties publicly comparing the basis selection results through a classical channel, retaining only the quantum events where the basis is consistent, and generating the original key bit stream.

[0018] Error correction specifically involves using rate-compatible LDPC codes to correct errors in the original key;

[0019] Privacy amplification specifically involves using the SHA-3-256 hash function to perform hash operations on the error-corrected key, thereby compressing the key length.

[0020] When outputting the security key, parallel encoding and verification of the key are performed through a field-programmable gate array (FPGA), and LDPC encoding and hash operation are accelerated by a GPU cluster. Finally, the security key is output.

[0021] A signal processing device for a quantum key distribution system includes a signal sensing module, an adaptive processing module, a synchronization calibration module, and a key processing module. The signal sensing module is used to collect the time-domain and frequency-domain characteristics of quantum signals and noise and output the original electrical pulse sequence. The adaptive processing module uses a field-programmable gate array and a digital signal processor to suppress phase noise, polarization noise, and intensity noise. The synchronization calibration module is used to regulate and calibrate phase synchronization error and clock synchronization delay fluctuations. The key processing module uses a GPU cluster to perform post-processing and parallel computation on the key and outputs a secure key.

[0022] Further improvements include: the signal sensing module also includes an auxiliary detection module, which monitors the noise spectrum and polarization state distribution of the optical fiber channel in real time through the auxiliary detection channel and transmits the data to the adaptive processing unit; the field-programmable gate array includes a hardware acceleration module for accelerating the parallel computation of Wiener filtering, Kalman filtering and LSTM prediction.

[0023] The beneficial effects of this invention are as follows: This invention effectively improves the signal-to-noise ratio of quantum signals and reduces the bit error rate through multi-dimensional noise perception and adaptive Wiener filtering. It achieves high-precision phase and clock synchronization through closed-loop feedback control. It adopts a parallel computing architecture for calculation, which effectively improves the data processing efficiency and meets the real-time requirements of high-speed QKD systems. At the same time, the invention combines machine learning prediction and closed-loop feedback control, which can adaptively cope with the dynamic changes of optical fiber channels and improve robustness. Attached Figure Description

[0024] Figure 1 This is a flowchart of the method of the present invention.

[0025] Figure 2 This is a system architecture diagram of the device of the present invention. Detailed Implementation

[0026] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0027] Quantum key distribution (QKD) is based on fundamental principles of quantum mechanics (such as the quantum no-cloning theorem and the uncertainty principle), and can achieve unconditionally secure key distribution. It is one of the core technologies of future quantum communication networks.

[0028] In current mainstream QKD protocols (such as BB84, B92, and measurement device-independent QKD), signal processing is a key factor affecting system performance, but it suffers from the following problems:

[0029] Noise interference suppression: Spontaneous emission noise, nonlinear phase noise caused by the Kerr effect, and polarization drift caused by environmental vibration in the optical fiber channel can significantly reduce the signal-to-noise ratio (SNR) of single-photon detection, thereby increasing the bit error rate (BER).

[0030] Phase synchronization accuracy: Phase-coded QKD (such as phase-matching protocols) is highly sensitive to the phase consistency between the transmitter and receiver. In long-distance transmission, fiber phase drift (which can reach the rad level) will cause phase decoding errors.

[0031] Time synchronization error: Timestamp-based key screening requires precise synchronization of the clocks of the sending and receiving ends. Traditional synchronization methods (such as satellite time synchronization and fiber optic two-way time transmission) have problems such as delay fluctuations or complex deployment.

[0032] Data processing efficiency: High-speed QKD systems (such as those with transmission rates in the Gbps range) require signal processing units to have low latency and high parallel computing capabilities, and traditional software algorithms are difficult to meet real-time requirements.

[0033] In existing technologies, the following solutions are generally adopted to address the above problems, but they also have limitations:

[0034] Passive noise suppression using narrowband filtering to address noise interference sacrifices photon utilization; active phase compensation using feedback control addresses phase drift, but suffers from response delay and cannot adapt to rapidly changing channels; time synchronization relies on external devices, increasing system complexity; and many systems employ general-purpose digital signal processing (DSP) frameworks for data processing and computation without optimization for quantum state characteristics.

[0035] Example

[0036] according to Figure 1 and Figure 2 As shown, this embodiment provides a signal processing method for a quantum key distribution system, including the following steps:

[0037] Step 1: Multi-dimensional signal sensing and preprocessing. Noise characteristics of the sensing quantum signal channel are collected through the auxiliary detection channel, including noise spectrum, phase noise power spectral density and polarization state distribution. Valid events are screened and timestamps are calibrated.

[0038] Specifically, noise characteristics of the sensing quantum signal channel are collected through the auxiliary detection channel, and quantum signal components are extracted based on wavelet transform and blind source separation algorithms. Threshold discrimination and basis vector comparison are performed on the original electrical pulse signal output by the single-photon detector to screen valid events, eliminate misjudged "vacuum pulses" and "multiphoton pulses", and use hardware phase-locked loop and bidirectional fiber time transfer technology to calibrate the timestamp error between the transmitting end and the receiving end.

[0039] The noise characteristics are perceived in real time by using a 1310nm wavelength continuous wave monitoring light at the receiving end to collect Rayleigh scattering noise in the fiber optic channel. Wavelet transform is used to perform time-frequency decomposition of the 1550nm quantum signal and noise, and the quantum signal components are separated by a blind source separation algorithm.

[0040] The calibration of timestamp error is specifically carried out by first deploying hardware phase-locked loops at the transmitting and receiving ends to generate local synchronization clocks. Then, the transmitting end periodically sends optical pulses with timestamps, and the receiving end records the pulse arrival time through a high-speed time-to-digital converter. Finally, bidirectional fiber optic time transfer technology is used to calculate and calibrate the clock offset, and the timestamp is calibrated according to the obtained offset. The timestamp error can be calibrated to ≤8ps.

[0041] Step 2: Adaptive noise suppression. Based on the perceived noise features, a Wiener filter and machine learning model are constructed to suppress phase noise, polarization noise, and intensity noise.

[0042] Phase noise suppression is based on an adaptive Wiener filter with dynamically adjustable center frequency and bandwidth designed based on real-time perceived phase noise power spectral density. Then, an LSTM neural network is introduced to predict short-term fluctuations in phase noise with a prediction step size of 10-100ms. The coefficients of the Wiener filter are then adjusted based on the prediction results. Long-time averaging compensation is used for low-frequency phase drift ≤100Hz, and narrowband notch filtering is used for high-frequency noise ≥1kHz.

[0043] The suppression of polarization noise is achieved by splitting the quantum signal into two orthogonal polarization components using a polarization beam splitter, monitoring the intensity ratio of the two components, optimizing the voltage parameters of the programmable polarization controller using a gradient descent algorithm, and finally minimizing the polarization extinction ratio loss. This can improve the polarization extinction ratio from 10dB to over 25dB.

[0044] Suppressing intensity noise involves applying a Kalman filter algorithm to the electrical pulse sequence of a single-photon detector and establishing state transition and observation equations. Then, the event sequence is smoothed through a prediction-correction process. An abnormal pulse judgment threshold is set to remove abnormal pulses. Specifically, the threshold can be set to count > 10 times the average value within three consecutive time windows. The resulting effective event retention rate can be ≥ 98%.

[0045] Step 3: Phase and clock synchronization calibration, using closed-loop feedback control to optimize and control phase synchronization error and clock synchronization delay fluctuations;

[0046] Specifically, a phase reference driven by a temperature-controlled crystal oscillator is set at the receiving end. By comparing the deviation between the actual phase of the quantum signal and the reference phase, the output of the local phase modulator is adjusted in real time using a PID controller to optimize the phase synchronization error, which can be controlled within π / 10. Then, at the transmitting end, time-stamped optical pulses are periodically sent. At the receiving end, the delay is calculated by combining the fiber group velocity dispersion model and the local clock offset is dynamically updated to optimize the clock synchronization delay fluctuation, which can make the clock synchronization delay fluctuation ≤10ps.

[0047] Phase synchronization is achieved by using a 10MHz reference clock provided by a temperature-controlled crystal oscillator (OCXO) at the receiver to drive a phase modulator (PM) to modulate the phase of the quantum signal; the phase difference of the interference fringes is monitored by a balanced beat detector (BHD). Will Input to the PID controller and adjust the voltage of PM until... It remains stable within ±π / 15.

[0048] Clock synchronization is achieved by the transmitter sending a timestamped light pulse (marked time t) every 1ms. d The receiver TDC records the pulse arrival time t. rThe theoretical delay t is calculated using the fiber group velocity dispersion (GVD) model v = c / n (n = 1.468). d =L / v, where L is the fiber length, and the final synchronized clock offset Δt = t r -(t d +Δt0)≤12ps, where Δt0 is the initial offset.

[0049] Step 4: Efficient key processing and post-processing. An architecture using basis vector comparison, low-density parity check code and SHA-3 hash function in parallel computation is used to extract the original key, perform error correction and privacy amplification, and output the secure key.

[0050] The original key extraction process involves both parties publicly comparing the basis selection results through a classical channel, retaining only the quantum events with consistent basis selection, and generating the original key bit stream, with consistent basis selection events accounting for ≥45%.

[0051] Error correction specifically involves using a rate-compatible LDPC code with a rate of 0.7-0.9 to correct errors in the original key. After correction, the bit error rate is reduced from 10. -2 The magnitude was reduced to ≤5×10 -4 ;

[0052] Privacy amplification specifically uses the SHA-3-256 hash function to perform hash operations on the error-corrected key, compressing the key length to 1 / 3 to 1 / 2 of the original length, ensuring that eavesdroppers cannot obtain the complete key information;

[0053] When outputting the security key, parallel encoding and verification of the key are performed through a field-programmable gate array (FPGA), such as parallel processing of 1024 bits / clock cycle. Combined with GPU cluster acceleration of LDPC encoding and hash operation, the security key is finally output, which can achieve high-speed output with an output rate of ≥8Gbps.

[0054] A signal processing device for a quantum key distribution system includes a signal sensing module, an adaptive processing module, a synchronization calibration module, and a key processing module.

[0055] The signal sensing module is used to collect the time-domain and frequency-domain characteristics of quantum signals and noise and output the original electrical pulse sequence. It integrates a 1310nm wavelength continuous wave monitor, a high-speed photodetector with a detection efficiency of ≥20%, and a time-to-digital converter with a time resolution of ≤50ps.

[0056] The adaptive processing module uses field-programmable gate arrays and digital signal processors to suppress phase noise, polarization noise and intensity noise. It is used to deploy noise suppression algorithms and phase / time synchronization control modules, and supports real-time parameter updates and algorithm iterations.

[0057] The synchronization calibration module is used to adjust and calibrate phase synchronization error and clock synchronization delay fluctuation, including frequency stability ≤10. -12 Thermostatic crystal oscillator, programmable phase modulator with voltage control accuracy ≤0.1V, and fiber delay line with delay adjustment range of 0-100ns;

[0058] The key processing module performs parallel computations on the key based on the GPU cluster, including LDPC encoding and privacy amplification, and outputs the security key at a rate of ≥8Gbps.

[0059] The signal sensing module also includes an auxiliary detection module, which monitors the noise spectrum and polarization state distribution of the optical fiber channel in real time through the auxiliary detection channel and transmits the data to the adaptive processing unit. The auxiliary detection channel is a continuous wave monitoring optical transmitter and receiver. The field programmable gate array contains a hardware acceleration module to accelerate the parallel computation of Wiener filtering, Kalman filtering and LSTM prediction.

[0060] Application examples

[0061] This application example uses the signal processing of a 100km fiber optic QKD system to illustrate the application process of the present invention.

[0062] Transmitter configuration: Continuous variable QKD (CV-QKD) light source (center wavelength 1550nm, linewidth 100kHz), using Mach-Zehnder modulator (MZM) for phase and amplitude modulation;

[0063] Receiver configuration: Balanced beat detector (BHD), time resolution 50ps, detection efficiency 20%;

[0064] Fiber optic channel configuration: standard single-mode fiber (SMF-28), loss 0.2dB / km, dispersion 17ps / (nm·km), ambient temperature fluctuation ±1℃ / h.

[0065] Signal processing flow:

[0066] 1. Multi-dimensional signal perception:

[0067] The auxiliary monitoring channel (wavelength 1310nm) collects Rayleigh scattering noise from the optical fiber in real time, and obtains the time-frequency distribution of the quantum signal (1550nm) and noise through wavelet transform decomposition.

[0068] The electrical pulses output by the single-photon detector are sampled by the TDC to generate a timestamp sequence (accuracy 50ps), which is synchronized with the transmitter via the TWSTFT, and the timestamp error is calibrated to 8ps.

[0069] 2. Adaptive noise suppression:

[0070] The phase noise (PSD) was obtained through FFT analysis, and its dominant frequency components were concentrated in the 10-100Hz range (low-frequency drift) and the 1-10kHz range (Kerr noise). The adaptive Wiener filter dynamically adjusted the filter order and cutoff frequency based on the PSD, using long-time averaging compensation for low-frequency drift and narrowband notch filtering for high-frequency noise.

[0071] Polarization noise is measured by monitoring the light intensity ratio (IBHD) at the two output ports of the BHD. + / I - By using a PID controller to adjust the voltage of the PC, the polarization extinction ratio was increased from 10dB to 25dB.

[0072] Intensity noise is smoothed using a Kalman filter, and an abnormal pulse count >10 within three consecutive time windows is removed, resulting in an effective event retention rate of 98%.

[0073] 3. Phase synchronized with clock:

[0074] Phase synchronization: The receiver OCXO provides a 10MHz reference clock to drive the PM to modulate the phase of the quantum signal. The phase difference between the interference fringes output by the BHD and the ideal fringes is then compared. The PID controller adjusts the voltage of PM to make... It remains stable within ±π / 15.

[0075] Clock synchronization: The transmitter sends a timestamped light pulse t every 1ms. d The receiver TDC records the pulse arrival time t. r The theoretical delay t is calculated using the fiber group velocity v = c / n (n = 1.468). d =L / v (L=100km, t d ≈505μs), the final synchronized clock offset Δt=t r -(t d +Δt0)≤12ps, where Δt0 is the initial offset.

[0076] 4. Key processing and post-processing:

[0077] Original key extraction: After basis comparison, the proportion of events in which both parties choose the same basis is 48% (theoretical value 50%, slightly reduced due to noise);

[0078] Error correction: Using rate-compatible LDPC code (code rate 0.8), the bit error rate is reduced to 5×10 after error correction. -4 ;

[0079] Privacy Enhancement: The SHA-3-256 hash function is used to compress the key length to 1 / 3 of the original length, resulting in a final secure key rate of 8.2Gbps.

[0080] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A signal processing method for a quantum key distribution system, characterized in that, Includes the following steps: Step 1: Multi-dimensional signal perception and preprocessing. Noise characteristics of the sensing quantum signal channel are collected through the auxiliary detection channel to screen valid events and calibrate timestamps. Specifically, noise characteristics of the sensing quantum signal channel are collected through the auxiliary detection channel, and quantum signal components are extracted based on wavelet transform and blind source separation algorithms. Threshold discrimination and basis vector comparison are performed on the original electrical pulse signal output by the single-photon detector to screen valid events, eliminate misjudged "vacuum pulses" and "multiphoton pulses", and use hardware phase-locked loop and bidirectional fiber time transfer technology to calibrate the timestamp error between the transmitting end and the receiving end. Step 2: Adaptive noise suppression. Based on the perceived noise features, a Wiener filter and machine learning model are constructed to suppress phase noise, polarization noise, and intensity noise. Phase noise suppression is based on an adaptive Wiener filter with dynamically adjustable center frequency and bandwidth designed based on real-time perceived phase noise power spectral density. Then, an LSTM neural network is introduced to predict short-term fluctuations in phase noise with a prediction step size of 10-100ms. The coefficients of the Wiener filter are then adjusted based on the prediction results. Long-time averaging compensation is used for low-frequency phase drift ≤100Hz, and narrowband notch filtering is used for high-frequency noise ≥1kHz. The suppression of polarization noise is achieved by splitting the quantum signal into two orthogonal polarization components using a polarization beam splitter, monitoring the intensity ratio of the two components, optimizing the voltage parameters of the programmable polarization controller using a gradient descent algorithm, and finally minimizing the polarization extinction ratio loss. This can improve the polarization extinction ratio from 10dB to over 25dB. Suppressing intensity noise involves applying a Kalman filter algorithm to the electrical pulse sequence of a single-photon detector and establishing state transition and observation equations. Then, the event sequence is smoothed through a prediction-correction process. An abnormal pulse judgment threshold is set to remove abnormal pulses. Specifically, the threshold can be set to counts > 10 times the average value within three consecutive time windows. The resulting effective event retention rate can be ≥ 98%. Step 3: Phase and clock synchronization calibration, using closed-loop feedback control to optimize and control phase synchronization error and clock synchronization delay fluctuations; Step 4: Efficient key processing and post-processing. An architecture using basis vector comparison, low-density parity check code and SHA-3 hash function parallel computation is used to extract the original key, correct errors and amplify privacy before outputting the secure key.

2. The signal processing method for a quantum key distribution system according to claim 1, characterized in that: The noise characteristics are specifically perceived by acquiring Rayleigh scattering noise of the fiber optic channel in real time through continuous wave monitoring light at the receiving end, using wavelet transform to perform time-frequency decomposition of quantum signal and noise, and combining blind source separation algorithm to separate quantum signal components.

3. The signal processing method for a quantum key distribution system according to claim 1, characterized in that: The calibration of the timestamp error specifically involves first deploying hardware phase-locked loops at both the transmitting and receiving ends to generate local synchronization clocks. Then, the transmitting end periodically sends optical pulses with timestamps, and the receiving end records the pulse arrival time using a high-speed time-to-digital converter. Finally, bidirectional fiber optic time transfer technology is used to calculate and calibrate the clock offset, and the timestamp is calibrated based on the obtained offset.

4. The signal processing method for a quantum key distribution system according to claim 1, characterized in that: In step three, a phase reference driven by a temperature-controlled crystal oscillator is set at the receiving end. By comparing the deviation between the actual phase of the quantum signal and the reference phase, the output of the local phase modulator is adjusted in real time using a PID controller to optimize the phase synchronization error. Then, the transmitting end periodically sends time-stamped optical pulses, and the receiving end calculates the delay and dynamically updates the local clock offset by combining the fiber group velocity dispersion model to optimize the clock synchronization delay fluctuation.

5. The signal processing method for a quantum key distribution system according to claim 1, characterized in that: In step four, the extraction of the original key specifically involves both parties publicly comparing the basis selection results through a classical channel, retaining only quantum events where the basis is consistent, and generating the original key bit stream. Error correction specifically involves using rate-compatible LDPC codes to correct errors in the original key; Privacy amplification specifically involves using the SHA-3-256 hash function to perform hash operations on the error-corrected key, thereby compressing the key length. When outputting the security key, parallel encoding and verification of the key are performed through a field-programmable gate array (FPGA), and LDPC encoding and hash operation are accelerated by a GPU cluster. Finally, the security key is output.

6. The signal processing apparatus for a quantum key distribution system according to any one of claims 1-5, characterized in that: The system includes a signal sensing module, an adaptive processing module, a synchronization calibration module, and a key processing module. The signal sensing module is used to collect the time-domain and frequency-domain characteristics of quantum signals and noise and output the original electrical pulse sequence. The adaptive processing module uses a field-programmable gate array and a digital signal processor to suppress phase noise, polarization noise, and intensity noise. The synchronization calibration module is used to regulate and calibrate phase synchronization error and clock synchronization delay fluctuations. The key processing module uses a GPU cluster to perform post-processing and parallel computation on the key and outputs a secure key.

7. The signal processing device for a quantum key distribution system according to claim 6, characterized in that: The signal sensing module also includes an auxiliary detection module, which monitors the noise spectrum and polarization state distribution of the optical fiber channel in real time through the auxiliary detection channel and transmits the data to the adaptive processing unit. The field-programmable gate array includes a hardware acceleration module for accelerating the parallel computation of Wiener filtering, Kalman filtering, and LSTM prediction.

Images