Satellite non-area remote sensing data backhaul method based on single photon detection

Abstract

This application relates to the field of satellite optical communication and remote sensing data transmission technology, and discloses a satellite non-domain remote sensing data backhaul method based on single-photon detection. This method generates a two-dimensional spatially encoded signal light field and an orthogonally polarized point-source reference pilot light field based on binary data at the satellite transmitter. These two fields are spatiotemporally combined and transmitted to a scattering channel. At the ground receiver, the light field is separated according to polarization state, and photons are collected using a single-photon array. An asymmetric time window is set based on the time statistics of the reference light field to filter the photon-generated signal and the reference speckle image. The channel point spread function is estimated using the reference speckle, and a deconvolution operation is performed on the signal speckle in the frequency domain to reconstruct the spatial pattern. The data is then demodulated by correlation matching with a standard basis pattern. This invention effectively suppresses diffuse noise and eliminates channel ambiguity through orthogonal pilot co-path transmission and time-domain gating, achieving highly robust data backhaul under strong scattering environments.

Description

Technical Field

[0001] This invention relates to the field of satellite optical communication and remote sensing data transmission technology, specifically to a method for satellite non-domain remote sensing data backhaul based on single-photon detection. Background Technology

[0002] With the continuous improvement of Earth observation resolution, the amount of data generated by satellite remote sensing technology is growing exponentially, placing extremely high demands on the bandwidth of the satellite-to-ground data backhaul link. Free-space optical communication, with its advantages of high carrier frequency, fast transmission rate, abundant spectrum resources, and strong resistance to electromagnetic interference, has become an important means to solve the bottleneck of massive remote sensing data backhaul. However, the optical communication link between satellites and ground stations is highly susceptible to the influence of complex atmospheric environments, especially the presence of scattering media such as clouds, dense fog, and dust, which often block the direct line-of-sight transmission path, creating non-line-of-sight communication scenarios. This poses a significant challenge to the reliable transmission of optical signals.

[0003] In non-line-of-sight scattering channels, photons undergo multiple random collisions with medium particles during transmission, leading to severe distortion of the wavefront phase and disruption of spatial coherence. Simultaneously, photons from different paths arrive at the receiver at varying times, causing pulse broadening in the time domain and resulting in severe multipath effects and inter-symbol interference. Traditional intensity-modulated / direct-detection-based satellite optical communication systems are highly dependent on the directional transmission and wavefront quality of the beam. In strong scattering environments, the receiver's light spot experiences severe speckle fragmentation and energy attenuation, causing a sharp increase in the bit error rate or even interruption of the communication link. While adaptive optics can compensate for phase disturbances caused by atmospheric turbulence through wavefront detection and correction, it lacks effective compensation capabilities for strong scattering and multipath effects caused by volumetric scattering media such as clouds and fog. Furthermore, these systems are often bulky and costly, making it difficult to maintain effectiveness over a large field of view. In addition, while some existing non-line-of-sight imaging techniques can reconstruct hidden targets using scattered light, they typically employ active scanning modes, resulting in slow data acquisition speeds that fail to meet the real-time requirements of high-speed satellite downlink data transmission. Therefore, how to achieve robust satellite optical data transmission under harsh channel conditions of strong scattering and non-line-of-sight is a key technical challenge that urgently needs to be solved in this field. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a satellite non-line-of-sight remote sensing data backhaul method based on single-photon detection. This method solves the problem that existing satellite optical communication systems cannot achieve highly reliable data backhaul under non-line-of-sight strong scattering channel conditions such as cloud cover and fog, where atmospheric multiple scattering causes severe spatial blurring of the signal light field and temporal multipath broadening.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: The present invention is a satellite non-domain remote sensing data backhaul method based on single-photon detection. This method is designed for communication scenarios where there is a scattering medium blocking the communication between the satellite and the ground station or there is no direct line of sight. By constructing an orthogonal polarization common path optical field at the transmitting end and combining time gating and speckle deconvolution technology at the receiving end, reliable data transmission and demodulation can be achieved.

[0006] Specifically, this method first constructs the optical field at the satellite transmitter. Based on the binary data to be transmitted, a first linearly polarized signal optical field carrying a two-dimensional spatially encoded pattern is generated; simultaneously, a second linearly polarized reference pilot optical field with a spatial distribution approximately resembling a point source is generated, and this second linearly polarized state is orthogonal to the first linearly polarized state. Subsequently, the signal optical field and the reference pilot optical field are aligned and combined spatially and temporally to generate an orthogonally polarized common-path composite beam. After this composite beam is emitted into the atmospheric scattering transmission channel, the signal optical field and the reference pilot optical field experience the same scattering path, thus ensuring that the channel distortion experienced by both has a high spatial correlation, i.e., satisfying the optical memory effect.

[0007] At the ground receiver, the received scattered light field is separated into a first component corresponding to the signal light field and a second component corresponding to the reference pilot light field through a polarization beam splitting mechanism. Since scattering causes the photon arrival distribution to broaden over time, this method uses a single-photon detector array in conjunction with time-correlated single-photon counting technology to collect the photon arrival events of the two components separately.

[0008] To suppress diffuse noise caused by multipath effects, this method introduces an asymmetric time window filtering mechanism. Based on the statistical distribution of photon arrival times in the reference pilot light field (second component), the average arrival time of quasi-ballistic photons is determined. An asymmetric time window is set based on this time, where the backward truncation time span is smaller than the trailing edge time span of the full width at half maximum (FWHM) of the scattered pulse. This time window is used to perform temporal filtering on the acquired photon arrival events, retaining only early-arriving photons carrying effective spatial information, and accumulating them to generate a two-dimensional signal speckle image and a two-dimensional reference speckle image. This process significantly improves the signal-to-noise ratio of the received speckle image.

[0009] The data recovery phase employs a strategy combining frequency-domain deconvolution and correlation matching. A two-dimensional reference speckle image is used as a real-time estimate of the channel point spread function at the current moment. A Wiener filter is constructed in the frequency domain, and deconvolution is performed on the two-dimensional signal speckle image. This operation eliminates image blurring caused by channel scattering and reconstructs the spatial coding pattern. Finally, by calculating the correlation between the reconstructed spatial coding pattern and the standard basis pattern in the preset codebook, the original binary data is demodulated using the maximum correlation decision, completing the data backhaul of the communication link.

[0010] In a specific transmitter implementation, the laser pulse sequence generated by the pulsed laser source is divided into two paths. The first optical path uses a spatial light modulator to load a two-dimensional spatially encoded pattern and adjust the polarization state; the second optical path uses a spatial filter to filter out higher-order transverse modes to form a point source light field and adjust the polarization state. The two paths are combined by a polarization beam combiner to ensure that the pulse envelopes coincide on the time axis and the principal optical axes coincide on the spatial axis, thereby achieving common-path detection of the scattering channel.

[0011] In one specific data acquisition implementation, the system generates a four-dimensional photon event data stream containing spatial coordinates, absolute arrival time, and polarization channel identifiers. The data stream is divided into a signal photon set and a reference photon set using the polarization channel identifiers. By statistically analyzing the time-of-flight histogram of the reference photon set and identifying peak times, the reference position of the asymmetric time window is determined, thereby accurately eliminating long-path diffuse photons.

[0012] In a specific deconvolution implementation, two-dimensional Fourier transforms are performed on both the two-dimensional signal speckle image and the two-dimensional reference speckle image to obtain the spectrum. The construction of the Wiener filter involves multiplying the signal spectrum by the complex conjugate of the reference spectrum to obtain the numerator, and calculating the square of the modulus of the reference spectrum and adding it to a regularization parameter to obtain the denominator. The reconstructed spectrum is obtained through division between the numerator and denominator, where the regularization parameter suppresses numerical instability at spectral zeros. After inverse two-dimensional Fourier transform and modulus taking, the reconstructed spectrum is restored to its spatially coded pattern.

[0013] In one specific decoding implementation, to eliminate the influence of received light intensity fluctuations, the reconstructed pattern and the standard base pattern are mean-centered. The Pearson correlation coefficient is calculated through dot product and normalization, and the binary data sequence is mapped back using a lookup table based on the maximum value index of the Pearson correlation coefficient. The two-dimensional spatial coding pattern can be either a Hadamard matrix or a random speckle pattern, and is updated periodically according to the data frame.

[0014] This invention provides a method for satellite non-domain remote sensing data backhaul based on single-photon detection. It has the following beneficial effects: 1. This invention generates an approximate point source reference pilot light field orthogonal to the signal light field at the transmitting end and adopts a common-path beam combining transmission method. This ensures that the signal light and the reference light undergo a highly correlated wavefront distortion process in the atmospheric scattering channel. This allows the receiving end to estimate the channel point spread function in real time and accurately using the speckle distribution of the reference pilot light field. Furthermore, the spatial ambiguity effect caused by atmospheric turbulence and cloud scattering can be effectively offset by deconvolution operation. This allows for the reconstruction of the light field under dynamic scattering environment without the need for a complex adaptive optics system.

[0015] 2. This invention utilizes the statistical distribution of photon arrival time in the reference pilot light field to set an asymmetric time window, achieving precise temporal filtering of the received photon stream. By filtering out delayed long-path diffuse photons and retaining quasi-ballistic photons carrying effective spatial information, it reduces signal broadening and background noise interference caused by multipath effects, improves the contrast and signal-to-noise ratio of the accumulated speckle image at the receiver, thereby extending the communication distance and reliability of the satellite remote sensing data backhaul link under non-line-of-sight or strong scattering conditions.

[0016] 3. This invention employs a decoding strategy combining frequency-domain Wiener filtering deconvolution with Pearson correlation matching, which effectively overcomes the problem of traditional direct imaging being sensitive to channel noise. Wiener filtering suppresses high-frequency noise amplification during the spectrum recovery process through regularization parameters, while the correlation-based decision mechanism eliminates the influence of light intensity fluctuations caused by atmospheric scintillation on data decision-making, ensuring that a low bit error rate can still be maintained even in low photon flux and high bit error rate environments, thus achieving highly robust non-domain data demodulation. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the algorithm data flow for the speckle self-deconvolution and light field reconstruction process of the present invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0019] The present invention first includes a satellite non-domain remote sensing data backhaul system based on single-photon detection, which includes a satellite transmission subsystem, an atmospheric scattering transmission channel, and a ground receiving subsystem.

[0020] The satellite launch subsystem generates and transmits a signal beam carrying remote sensing data and a reference pilot beam for channel sounding. The ground receiving subsystem receives the photon stream after it has passed through the atmospheric scattering transmission channel and performs data demodulation.

[0021] The satellite launch subsystem includes a pulsed laser source, a beam splitter module, a signal modulation optical path, a reference pilot optical path, a beam combiner module, and a transmitting optical antenna.

[0022] A pulsed laser source is used to generate a high-repetition-rate time-synchronized laser pulse sequence. This pulsed laser source is connected to a beam-splitting module via optical fiber or a free-space optical path. The beam-splitting module spatially separates the incident laser pulse sequence into a first beam path and a second beam path.

[0023] The signal modulation optical path is positioned on the first beam path. The signal modulation optical path includes a spatial light modulator and a first polarization controller.

[0024] The spatial light modulator is electrically connected to the satellite's data storage unit to receive the binary remote sensing data stream to be transmitted back. Based on the received binary data, the spatial light modulator loads the corresponding two-dimensional spatially encoded pattern onto the optical field cross-section of the first beam path.

[0025] The first polarization controller is used to adjust the beam modulated by the spatial light modulator to a first linear polarization state, thereby forming a signal light field.

[0026] The reference pilot optical path is positioned on the second beam path. The reference pilot optical path includes a point source generator and a second polarization controller.

[0027] The point source generator is used to spatially filter the beam in the second beam path to generate a point source light field or a plane wave light field without spatial encoding information.

[0028] The second polarization controller is used to adjust the point source light field to the second linear polarization state, thereby forming a reference pilot light field.

[0029] The first linearly polarized state and the second linearly polarized state are orthogonal to each other in polarization direction.

[0030] The beam combining module includes a polarization beam combiner. The first input port of the polarization beam combiner is coupled to the output of the signal modulation optical path, and the second input port is coupled to the output of the reference pilot optical path.

[0031] A polarization beam combiner is used to spatially coincide a signal light field with a first linear polarization state with a reference pilot light field with a second linear polarization state, and to maintain pulse synchronization in time, thereby forming an orthogonally polarized common-path composite beam.

[0032] The transmitting optical antenna is coupled to the output of the beam combiner module to direct the orthogonally polarized common-path combined beam to the atmospheric scattering transmission channel.

[0033] The atmospheric scattering transmission channel is located between the satellite launch subsystem and the ground receiving subsystem. This channel contains media with random scattering characteristics, including clouds, aerosols, or regions of atmospheric turbulence.

[0034] The ground receiving subsystem includes a receiving optical antenna, a polarization beam splitter module, a signal detection array, a reference detection array, a time-correlation counting module, and a data processing unit.

[0035] The receiving optical antenna is used to collect the scattered light field passing through the atmospheric scattering transmission channel and guide the collected light field to the polarization beam splitter module.

[0036] The polarization beam splitter module is used to separate the incident scattered light field into a first component beam and a second component beam according to the polarization state. The first component beam corresponds to the signal light field at the transmitting end, and the second component beam corresponds to the reference pilot light field at the transmitting end.

[0037] The signal detection array is located on the imaging path of the first component beam. The signal detection array consists of multiple single-photon avalanche diode pixel units arranged in a matrix.

[0038] The reference detector array is located in the imaging path of the second component beam. The reference detector array consists of multiple single-photon avalanche diode pixel units arranged in a matrix.

[0039] Both the signal detection array and the reference detection array are electrically connected to the time-correlated counting module.

[0040] The time-correlated counting module is used to record the moment when each pixel unit in the signal detection array and the reference detection array generates a photon avalanche event, and generates a raw photon event data stream containing spatial coordinate information, time information and polarization channel identification information.

[0041] The data processing unit is electrically connected to the time-correlated counting module and is used to receive the raw photon event data stream.

[0042] The data processing unit includes a memory and a processor. The memory stores computer-executable instructions. When the instructions are executed by the processor, the processor performs time-domain filtering, frequency-domain cross-correlation deconvolution operations, and correlation matching decoding on the raw photon event data stream to output demodulated binary data.

[0043] See attached document Figure 1 This invention provides a method for satellite non-domain remote sensing data backhaul based on single-photon detection, the method comprising the following steps:

[0044] Generating an orthogonally polarized common-path composite beam: At the satellite transmitter, a high-repetition-rate time-synchronized laser pulse is generated. This laser pulse is divided into a first optical path and a second optical path. Binary remote sensing data to be transmitted is acquired and mapped into a two-dimensional spatial coding pattern selected from a preset set of orthogonal bases. This two-dimensional spatial coding pattern is loaded into the optical field of the first optical path using a spatial light modulator, and the optical field is adjusted to a first linear polarization state to form a signal optical field. Simultaneously, the optical field of the second optical path is spatially filtered to form a point source optical field or a plane wave optical field, and this optical field is adjusted to a second linear polarization state orthogonal to the first linear polarization state to form a reference pilot optical field. Subsequently, the signal optical field and the reference pilot optical field are combined so that they coincide in the spatial cross-section and are aligned in the time pulse, generating an orthogonally polarized common-path composite beam.

[0045] Transmission via scattering channel: The orthogonally polarized co-path synthesized beam is emitted into the atmospheric scattering transmission channel. The beam undergoes multiple scattering and wavefront distortion as it passes through clouds, aerosols, or atmospheric turbulence. Because the signal light field and the reference pilot light field have spatial co-path characteristics and are located within the same isohalo region, their instantaneous point spread function and phase perturbation during transmission remain consistent.

[0046] Acquiring four-dimensional photon event data: At the ground receiver, a large-aperture optical antenna collects the scattered and transmitted light field. A polarization beam splitter separates the collected light field into a first component corresponding to the signal light field and a second component corresponding to the reference pilot light field. A first single-photon detector array and a second single-photon detector array are used to detect photons in these two components, respectively. Using time-correlated single-photon counting technology, the spatial coordinates, absolute arrival time, and polarization channel identifier of each arriving photon event are asynchronously recorded to generate the original photon event data stream.

[0047] Temporal gating and speckle image generation: Temporal analysis is performed on the raw photon event data stream. The photon arrival time distribution corresponding to the reference pilot light field is statistically analyzed to determine the peak arrival time of quasi-ballistic photons. An asymmetric time window is set based on this peak time, covering the arrival periods of quasi-ballistic photons and serpentine photons, and truncating the long-tailed diffuse photon periods. The photon data of the signal light field and the reference pilot light field are filtered using this asymmetric time window to remove background noise photons and multipath scattered photons outside the time window. The filtered photons are accumulated in the spatial domain to generate two-dimensional signal speckle images and two-dimensional reference speckle images, respectively.

[0048] Speckle self-deconvolution optical field reconstruction: Two-dimensional Fourier transforms are performed on both the two-dimensional signal speckle image and the two-dimensional reference speckle image to obtain their respective spatial spectral distributions. Using the spectrum of the reference speckle image as an estimate of the channel transfer function, cross-correlation deconvolution or Wiener filtering is performed on the spectrum of the signal speckle image to eliminate phase distortion and blurring caused by channel scattering, obtaining the reconstructed signal spectrum. An inverse Fourier transform is then performed on the reconstructed signal spectrum to obtain the reconstructed spatial coding pattern.

[0049] Correlation matching decoding: The reconstructed spatial coding pattern is compared with a pre-stored coding codebook at the receiver. This codebook contains all preset orthogonal basis patterns used by the transmitter. The correlation coefficient between the reconstructed spatial coding pattern and each basis pattern in the codebook is calculated. The index value corresponding to the basis pattern with the highest correlation coefficient is selected. According to a preset mapping relationship, this index value is converted into corresponding binary data, completing the demodulation process for data transmission back.

[0050] In this embodiment, the optical field construction process at the transmitter mainly includes three sub-steps: data mapping, dual-channel optical field modulation, and polarization co-path synthesis.

[0051] First, the data processing module of the satellite launch subsystem receives the binary data stream to be transmitted back. The data processing module divides the continuous binary data stream into data frames of fixed length. For each data frame, the data processing module converts it into a corresponding two-dimensional spatial matrix according to a preset mapping rule. This mapping rule is based on a predefined set of orthogonal basis functions. In this embodiment, the set of orthogonal basis functions is either the Hadamard matrix set or a random speckle pattern set. Let the current time... The corresponding data frame is mapped as a signal light field in a two-dimensional spatial pattern. ,in This represents the spatial coordinates of a cross section perpendicular to the direction of light propagation.

[0052] Next, a pulsed laser source generates a laser with a specific repetition frequency. and pulse width The original laser pulse sequence. This original laser pulse sequence is split into a first beam and a second beam with equal energy by a beam splitter.

[0053] The first beam enters the signal modulation optical path. A spatial light modulator is installed in the signal modulation optical path. The spatial light modulator receives the signal light field of the aforementioned two-dimensional spatial pattern. The driving signal is used to modulate the wavefront of the first beam splitter. The modulation method can be amplitude modulation or phase modulation, causing the cross-sectional intensity distribution or phase distribution of the first beam splitter to carry... The modulated beam, after passing through a first polarizer or polarization rotator, is set to a first linear polarization state (e.g., horizontal polarization state). This forms a signal light field. .

[0054] The second beam enters the reference pilot optical path. A spatial filter, such as a pinhole aperture or a single-mode fiber coupler, is installed in the reference pilot optical path. The spatial filter is used to filter out the higher-order transverse modes of the second beam, making its spatial distribution approximately that of an ideal point source or plane wave. Its spatial distribution function can be expressed using the Dirac delta function. function Approximate representation. The spatially filtered beam, after passing through a second polarizer or polarization rotator, is set to a second linear polarization state (e.g., vertical polarization state). The second linear polarization state is orthogonal to the first linear polarization state. This forms a reference pilot optical field. .

[0055] Subsequently, the signal light field and reference pilot light field Both beams enter the polarization combiner. The polarization combiner has the optical characteristic of transmitting horizontally polarized light and reflecting vertically polarized light (or vice versa). By adjusting the optical path difference between the signal modulation optical path and the reference pilot optical path, the two beams are made to satisfy the time synchronization condition when they arrive at the polarization combiner, that is, the time envelopes of the light pulses coincide on the time axis. At the same time, the collimating device is adjusted so that the two beams satisfy the spatial coaxial condition at the output of the polarization combiner, that is, the principal optical axes of the two beams coincide. After beam combining, the generated total emitted light field It contains two orthogonal polarization components, and its mathematical expression is as follows: ; in, The time envelope function of a laser pulse. The unit vector representing the first linear polarization state. The unit vector representing the second linear polarization state.

[0056] Through the above steps, the system constructs a composite optical field containing data probe pairs at the physical level. The signal component in this composite optical field carries valid information, while the reference component carries baseline information for probing the channel state. Because the two components are highly spatially overlapping and temporally synchronized, a physical basis is provided for subsequently utilizing optical memory effects at the receiver to eliminate scattering effects.

[0057] This embodiment illustrates the physical transformation mechanism that an orthogonally polarized common-path synthesized beam undergoes during its transmission from the satellite transmitter to the ground receiver.

[0058] After the orthogonally polarized co-path synthesized beam leaves the satellite's transmission aperture, it enters a non-free-space transmission channel. This channel contains either an atmospheric turbulent layer with a random refractive index distribution or a cloud layer with a high concentration of scattering particles. As it passes through this scattering medium, photons undergo multiple scattering events, resulting in disruption of the spatial coherence of the wavefront phase and a discretization broadening of the photon arrival time.

[0059] For scattering media, within the isoplanatic patch defined by the optical memory effect, the transmission of light through the medium can be approximated as a linear shift system over a short period. This is due to the signal light field at the transmitting end... With reference pilot light field After precise optical axis calibration, the two are highly overlapping in space, and the field of view of the signal light field is limited to the memory effect angle range of the medium. Therefore, although the first linear polarization state of the signal light field and the second linear polarization state of the reference pilot light field are orthogonal to each other, when passing through the same scattering volume, the instantaneous point spread function (PSF) experienced by the two remains highly consistent in statistical characteristics.

[0060] The instantaneous point spread function is denoted as It characterizes the scattering channel at time... The spatial response (i.e., speckle pattern) produced by an ideal point light source.

[0061] When the synthesized beam reaches the receiving plane, due to wavefront mixing caused by multiple scattering, the received light field no longer presents the original pattern at the transmitting end, but rather a seemingly randomly distributed granular speckle pattern. According to the convolution principle of linear optical systems, the instantaneous light intensity distribution on the receiving plane is determined by the two-dimensional spatial convolution of the emitted light field and the channel point spread function.

[0062] Specifically, the intensity distribution of the signal light component corresponding to the first linear polarization state on the receiving surface The signal light field represented as the original data pattern With channel point spread function Convolution: ; Intensity distribution of the reference pilot light component corresponding to the second linear polarization state at the receiving surface The reference pilot light field is represented as an ideal point source. With channel point spread function The convolution of the Dirac function. According to the properties of convolution, the result of the convolution of the Dirac function is the function itself; therefore, the intensity distribution of the reference light component directly reflects the current channel state. ; in, This represents a two-dimensional spatial convolution operation. This represents the pulse time envelope that undergoes time-domain broadening after transmission through the channel. Through the above physical process, the originally unknown channel interference information is completely copied into the speckle pattern of the reference pilot light component. Since the signal light component and the reference light component are transmitted simultaneously (on the same path), Captured in real time Every tiny phase distortion and scattering disturbance it experiences provides a precise physical reference for subsequent mathematical operations at the receiver to remove channel effects.

[0063] In this embodiment, the digital acquisition process of the scattered light field by the receiving subsystem is completed by a parallel dual-channel single-photon detector array in conjunction with a high-precision time measurement unit.

[0064] The ground station's receiving optical antenna collects and collimates the scattered beam into a parallel beam. This parallel beam is then incident on a polarization beam splitter module. The polarization beam splitter module includes a polarization beam splitter prism whose optical axis is aligned with the polarization state of the transmitter. The polarization beam splitter prism physically separates the incident beam into a first beam and a second beam. The first beam contains signal photons with a first linear polarization state, and the second beam contains reference pilot photons with a second linear polarization state.

[0065] The first beam is projected onto the photosensitive surface of the signal detection array, and the second beam is projected onto the photosensitive surface of the reference detection array. Both the signal detection array and the reference detection array are two-dimensional Geiger-mode avalanche photodiode (SPAD) arrays. Each pixel unit in the array is independently connected to its own readout circuit and has single-photon-level response capability. When a photon strikes the avalanche region of a pixel unit, that pixel unit generates an avalanche current pulse.

[0066] The Time-Correlated Counting Module (TCSPC) is electrically connected to each pixel unit of the two detector arrays. The TCSPC module internally contains multiple Time-to-Digital Converter (TDC) channels. When an avalanche current pulse is received from the pixel unit, the TDC channels record the arrival time of the pulse's leading edge relative to the system synchronization clock. This system synchronization clock is frequency-locked to the repetition frequency of the satellite's transmitted pulses.

[0067] In this process, the receiving system does not perform simulated integration of the light intensity, but instead asynchronously records each detected photon event in an event-driven manner. For the th For each detected photon event, the system generates a data packet containing four-dimensional information. The data structure for this four-dimensional information is defined as follows: ; in, The two-dimensional discrete spatial coordinates of the pixel unit responding to photons on the detector array plane; This indicates the absolute time stamp of the photon's arrival, and its time resolution is determined by the quantization precision of the TDC. This is the polarization channel identifier, used to indicate whether the photon comes from the signal detector array (corresponding to the signal light component) or the reference detector array (corresponding to the reference light component).

[0068] The system is set to a single-frame acquisition cycle. All recorded photon events within this period constitute the raw photon stream dataset. The data processing unit then processes the data based on the polarization channel identifier bits. The value of the original photon stream dataset is used to divide it into two subsets: the signal photon set and the signal photon set. and reference photon set . signal photon collection Includes all that satisfy The photon event quadruple pointing to the signal channel physically corresponds to the discrete sampling of the received signal speckle field.

[0069] Reference photon set Includes all that satisfy The photon event quadruple pointing to the reference channel physically corresponds to discrete sampling of the received channel point spread function. Through the above acquisition process, the continuous analog light field is converted into a digital photon event stream with extremely high temporal and spatial resolution, preserving complete spatiotemporal statistics of the arrival of scattered photons.

[0070] In this embodiment, after receiving the raw photon stream data, the data processing unit performs time-of-flight (ToF) statistical filtering to extract effective signal photons from strong background noise and long-path scattered photons. This process mainly includes four sub-steps: time-domain statistical histogram construction, asymmetric time window parameter calculation, photon event screening, and two-dimensional cumulative imaging.

[0071] First, the data processing unit processes the reference photon set. Time stamps of all photon events Statistical analysis is performed. The system divides a complete data frame period into multiple tiny time bins, counts the number of photons falling into each time bin, and thus constructs a time-of-flight histogram for the reference channel. This histogram physically characterizes the temporal impulse response of the current scattering channel. Due to atmospheric multipath effects, the histogram exhibits an asymmetric pulse shape: it contains a steep rising edge (corresponding to quasi-ballistic photons), a peak, and a slowly decaying long tail (corresponding to multiple diffuse photons).

[0072] Next, the data processing unit analyzes the histogram to determine the time gating parameters. The processor first identifies the moment with the maximum value in the histogram using a peak-finding algorithm, and records this as the peak moment. This moment represents the average arrival time of the photon group reaching the detector via the shortest scattering path. Subsequently, the system defines an asymmetric time gating window based on this peak moment. The time span of this asymmetric time gating window is defined as follows: .

[0073] in, The forward cutoff threshold is used to accommodate quasi-ballistic photons at the rising edge of the pulse; The backward truncation threshold is used to accommodate snake photons that carry some spatial information.

[0074] In this embodiment, The value of is strictly limited to be smaller than the trailing edge of the full width at half maximum (FWHM) of the scattering pulse, ensuring that most diffuse photons that have traveled long distances and lost all spatial coherence are excluded from the window. Simultaneously, because the solar background light and detector dark counts are uniformly distributed along the time axis, this narrow time window significantly reduces the total amount of noise photons accumulated in the final image.

[0075] After determining the time window parameters, the data processing unit synchronously applies the window to the signal photon set. and reference photon set Because the signal light from the transmitting end is strictly synchronized with the reference light in time, the reference channel is determined. The same applies to the signal channel. The processor iterates through each photon data set in both collections. Determine its time stamp Does it meet the following conditions: ; A photon event is marked as a valid photon and retained only if it meets the above conditions; otherwise, the photon is considered noise or invalid scattering and is discarded. Finally, the system performs spatial domain integration on the filtered valid photon data. For the signal channel, the processor calculates the spatial coordinates of all retained photons. Mapped onto a two-dimensional pixel grid, the total number of photons falling at each pixel position is calculated to generate a two-dimensional signal speckle image matrix. Similarly, for the reference channel, the processor generates a two-dimensional reference speckle image matrix. These two matrices represent the input data that has been temporally cleaned and has an enhanced signal-to-noise ratio, and are used for subsequent optical field deconvolution operations.

[0076] See attached document Figure 2 In this embodiment, the data processing unit executes the core optical field inversion algorithm, which aims to decouple the original emission spatial pattern from the speckle field of the signal channel using the channel state information carried by the reference channel. This process is based on the convolution theorem for linear systems and is implemented in the spatial frequency domain through regularized deconvolution operations.

[0077] The data processing unit first retrieves the two-dimensional signal speckle image matrix that has been filtered in the time domain from the memory. and two-dimensional reference speckle image matrix These two matrices are considered as discretized spatial domain light intensity distribution functions.

[0078] processor and Two-dimensional discrete Fourier transforms (DFTs) are performed to transform the image information from the spatial domain to the spatial frequency domain. This transformation process generates the corresponding complex spectrum matrix.

[0079] set up Here, we have spatial frequency coordinates. The spectrum of the signal speckle image is represented as follows: The spectrum of the reference speckle image is represented as follows: According to the physical model, It is approximately equivalent to the point spread function of the channel, while Original coded pattern The convolution with the spread function at that point. According to the convolution theorem, convolution in the spatial domain corresponds to multiplication in the frequency domain. Therefore, in the ideal, noise-free case, the spectrum of the original pattern should be the quotient of the signal spectrum and the reference spectrum.

[0080] However, considering the detection noise and the attenuation of high-frequency components in the spectrum, direct division would lead to a sharp amplification of noise near the zero point. Therefore, the processor employs the Wiener Deconvolution algorithm to construct a deconvolution filter. The processor then calculates and reconstructs the spectrum. Its operational logic is shown in the following formula: ; in, Represents the reference spectrum The complex conjugate matrix. This represents the power spectral density of the reference spectrum. This is the regularization parameter, whose value is determined by the system's preset signal-to-noise ratio estimate. It is used to suppress numerical instability when the denominator approaches zero and to prevent excessive amplification of high-frequency noise. It is obtained by multiplying by the complex conjugate. The algorithm effectively cancels out the phase distortion introduced by the channel transfer function.

[0081] Obtain the reconstructed spectrum The processor then performs a two-dimensional inverse discrete Fourier transform (IDFT) on the data, converting it back to the spatial domain. Since the original emission pattern is a light intensity distribution, which is a non-negative function in the real domain, the processor takes the modulus (or real part) of the complex matrix generated by the inverse transform to obtain the reconstructed spatial light intensity distribution matrix. .

[0082] ; in, This represents the inverse Fourier transform operator.

[0083] The final generated This is the estimated value of the original coded pattern after removing scattering effects. This reconstructed image preserves the spatial coding structure features of the transmitter while eliminating diffusion and distortion caused by atmospheric turbulence and cloud scattering, providing a clear image foundation for subsequent feature recognition and data decoding.

[0084] In this embodiment, the data processing unit performs the final decoding step of the communication link, that is, mapping the reconstructed analog grayscale image in step five back to a discrete digital bitstream. This process mainly includes four processing steps: standard base library calling, similarity traversal calculation, maximum likelihood decision, and binary inverse mapping. First, a standard orthogonal base pattern library is pre-installed in the memory of the ground receiving subsystem. This library contains... A predefined standard two-dimensional base pattern ,in These are pattern indexes. These standard base patterns are completely identical in mathematical structure and spatial resolution to the coded patterns loaded by the space light modulator at the satellite transmitter (e.g., if the transmitter uses a 64th-order Hadamard matrix, the receiver stores the corresponding 64 base patterns).

[0085] The processor receives the reconstructed image matrix output from the light field reconstruction step. Then, the traversal matching procedure is started. Considering that the attenuation characteristics of the atmospheric channel will cause the overall light intensity of the received image to fluctuate relative to the standard pattern, and that residual background noise may introduce DC bias, this embodiment uses the Pearson correlation coefficient as a similarity metric for anti-interference.

[0086] The processor calculates and reconstructs the image one by one. With each standard pattern in the base library Correlation coefficient between The calculation process centers the mean of the image matrix, and its expression is as follows: ; in, Represents the reconstructed image matrix The arithmetic mean of the gray levels of all pixels in the image. Indicates the first A standard pattern matrix The arithmetic mean of the gray levels of all pixels in the matrix. The normalization term in the denominator eliminates the differences in light intensity amplitude scaling caused by channel loss, thus improving the correlation coefficient. The value range is strictly limited to the interval [-1, 1], and it only reflects the similarity of spatial texture structure.

[0087] Finish After the secondary correlation calculation, the processor obtains a result from... A correlation vector is formed by several correlation coefficients. The system then executes the maximum likelihood decision logic (Winner-Takes-All). The processor compares the values ​​in the correlation vector, retrieves the correlation coefficient with the largest value, and extracts its corresponding index value. : ; in, For the target evaluation index, the index value It is determined to be the most likely spatial coding pattern number transmitted by the transmitter at the current moment.

[0088] Finally, the processor consults a pre-defined index data mapping table to obtain the decimal index from the decision. Convert to the corresponding length The binary data sequence is then fed into a backend channel decoding module (such as an error correction codec) to complete the full physical layer demodulation of a single data frame. By repeatedly executing the above steps, the system achieves continuous remote sensing data stream backhaul.

Claims

1. A method for satellite non-domain remote sensing data backhaul based on single-photon detection, characterized in that, Includes the following steps: At the satellite launch end, a first linearly polarized state signal light field carrying a two-dimensional spatial coding pattern is generated based on binary data, and a second linearly polarized state reference pilot light field that is approximately a point source and orthogonal to the first linearly polarized state signal light field is generated simultaneously. The first linearly polarized signal light field and the second linearly polarized reference pilot light field are spatiotemporally combined to generate an orthogonally polarized common-path composite beam, which is then emitted into the atmospheric scattering transmission channel. At the ground receiving end, the scattered light field is separated into a first component corresponding to the first linearly polarized signal light field and a second component corresponding to the second linearly polarized reference pilot light field, and photon arrival events are collected using a single-photon detection array. Based on the statistical distribution of photon arrival time of the second component, an asymmetric time window is determined. The asymmetric time window is used to filter photon arrival events, and two-dimensional signal speckle image and two-dimensional reference speckle image are generated by accumulating them respectively. Using the two-dimensional reference speckle image as an estimate of the channel point spread function, a deconvolution operation is performed on the two-dimensional signal speckle image in the frequency domain to obtain the reconstructed spatial coding pattern; By calculating the correlation between the reconstructed spatial coding pattern and the standard base pattern in the preset codebook, binary data is demodulated based on the maximum correlation.

2. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 1, characterized in that, The steps for generating the orthogonally polarized common-path composite beam specifically include: The laser pulse sequence generated by the pulsed laser source is divided into a first optical path and a second optical path; In the first optical path, the two-dimensional spatial coding pattern is loaded using a spatial light modulator, and the polarization state is adjusted to form the first linearly polarized signal light field; In the second optical path, a spatial filter is used to filter out the higher-order transverse modes of the beam to form a point source optical field, and the polarization state is adjusted to form the second linearly polarized reference pilot optical field; A polarization beam combiner is used to combine the first linearly polarized signal light field with the second linearly polarized reference pilot light field, so that their pulse envelopes coincide on the time axis and their principal optical axes coincide on the spatial axis.

3. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 1, characterized in that, The specific steps for collecting photon arrival events include: The time-correlated single-photon counting module is used to record the photon avalanche signal generated by each pixel unit in the single-photon detection array; Generate a four-dimensional photon event data stream containing spatial coordinate information, absolute time of arrival information, and polarization channel identification information; Based on the polarization channel identification information, the four-dimensional photon event data stream is divided into a signal photon set and a reference photon set.

4. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 3, characterized in that, The specific steps for determining an asymmetric time window include: The time-of-flight histogram is calculated by statistically analyzing the absolute arrival times of all photon events in the reference photon set; Identify the peak time of the time-of-flight histogram, where the peak time characterizes the average arrival time of quasi-ballistic photons; The asymmetric time window is set based on the peak time, wherein the backward truncation time span of the asymmetric time window is less than the backward time span of the full width at half maximum (FWHM) of the scattering pulse, thus filtering out long-path diffuse photons.

5. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 4, characterized in that, The specific steps for generating a two-dimensional signal speckle image and a two-dimensional reference speckle image include: Traverse the set of signal photons and the set of reference photons; Valid photon events whose absolute arrival time falls within the asymmetric time window are retained, while noisy photon events whose absolute arrival time is outside the asymmetric time window are removed. The two-dimensional signal speckle image and the two-dimensional reference speckle image are obtained by mapping the retained effective photon events to a two-dimensional pixel grid and performing counting and accumulation calculations.

6. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 1, characterized in that, The specific steps for performing a deconvolution operation include: The signal spectrum is obtained by performing a two-dimensional Fourier transform on the two-dimensional signal speckle image, and the reference spectrum is obtained by performing a two-dimensional Fourier transform on the two-dimensional reference speckle image. A Wiener filter is constructed in the frequency domain using the reference spectrum; The reconstructed spectrum is obtained by filtering the signal spectrum using the Wiener filter. The reconstructed spatial coding pattern is obtained by performing a two-dimensional inverse Fourier transform on the reconstructed spectrum and taking the modulus value.

7. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 6, characterized in that, The operation of obtaining the reconstructed spectrum by filtering the signal spectrum using the Wiener filter includes: The numerator is obtained by multiplying the signal spectrum by the complex conjugate of the reference spectrum; The denominator term is obtained by taking the square of the reference spectrum modulo and adding it to a preset regularization parameter; The reconstructed spectrum is obtained by performing a division operation between the numerator and the denominator, wherein the regularization parameter is used to suppress numerical instability at the zero point of the spectrum.

8. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 1, characterized in that, The preset codebook contains multiple standard basis patterns consistent with the orthogonal basis set used by the satellite transmitter; The specific steps for calculating correlation include: The reconstructed spatial coding pattern and the standard base pattern are respectively subjected to mean centering. The Pearson correlation coefficient is calculated by performing a dot product and normalization on the reconstructed spatial coding pattern after centralization and each standard base pattern in the preset codebook.

9. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 8, characterized in that, The specific steps for demodulating binary data include: Select the maximum value among the Pearson correlation coefficients and obtain the standard base pattern index corresponding to the maximum value; The standard base pattern index is converted into a binary data sequence according to a preset mapping table between indexes and binary values.

10. The satellite non-domain remote sensing data backhaul method based on single-photon detection according to claim 1, characterized in that, The two-dimensional spatial coding pattern is selected from the Hadamard matrix set or the random speckle pattern set, and the satellite transmitter updates the two-dimensional spatial coding pattern once in each data frame transmission cycle.

Images