Integrated array visible light communication receiver based on large area light collection and light field redistribution, method and system

Abstract

The application relates to the technical field of optical wireless communication, and discloses an integrated array visible light communication receiver based on large-area light collection and light field redistribution, a receiving method and a system. A fluorescent antenna converts first waveband signal light incident at different positions and angles of a light receiving area into second waveband fluorescent signal light and outputs the second waveband fluorescent signal light from different positions of an out-light end face of a waveguide; a plurality of silicon photomultiplier tube detection units arranged along the end face form a plurality of receiving branches through independent analog channels and synchronous sampling, perform waveform level equalization and discrete multi-tone demodulation based on a neural network, and then perform weighted combination in a symbol domain according to channel quality estimated by a pilot to restore original data. The application helps to expand the range of available incident positions and angles, improve rate uniformity and link stability, and reduce the difficulty of light beam alignment.

Description

Technical Field

[0001] This application relates to the field of optical wireless communication technology, and in particular to an integrated array visible light communication receiving technology based on large-area light collection and light field redistribution. Background Technology

[0002] With the development of mobile IoT, underwater exploration, and new human-computer interaction applications, visible light communication has attracted attention in various practical scenarios. For example, in high-density indoor access scenarios, handheld terminals or mobile robots need to maintain a high-speed data link with the optical communication access point while moving; in underwater operation scenarios, submersibles or underwater robots need to transmit large amounts of observation data back to the docking end under conditions of swaying attitude and constantly changing relative positions; in scenarios such as vehicle-to-everything (V2X), drone communication, and extended reality devices, the relative positions and attitudes of the transmitting and receiving ends change frequently, making it difficult for the light beam to be accurately aligned with the receiving end at all times. The common requirement for receivers in the above scenarios is that they can still maintain high-speed and stable optical signal reception under conditions of incident light spot position shift and incident angle change. However, existing high-speed visible light receiving technologies are difficult to meet these requirements. On the one hand, the photosensitive surface of high-speed photodetectors is usually small, and the receiving structure that uses lenses to focus the light is mainly effective for light incident at small angles along the normal direction. When the incident position or angle changes, the focused light spot is prone to deviate from the effective photosensitive area, resulting in insufficient link stability. Existing technologies have not yet effectively coordinated the relationship between transmission rate, effective receiving area, and receiving field of view. On the other hand, while using a wavelength-converting planar waveguide with a large light-receiving area can convert and guide the incident light to the output end, if only a single detector unit receives the light, or if the outputs of multiple detector units are directly connected in parallel in the analog domain, it is still difficult to retain and utilize the signal differences corresponding to different output positions. Furthermore, the nonlinearity of highly sensitive detectors such as silicon photomultiplier tubes can affect the quality evaluation and merging processing of multiple received signals. Therefore, there is an urgent need for a visible light communication receiving technology that can expand the range of usable incident positions and incident angles without increasing the photosensitive area of ​​a single high-speed detector unit, and can effectively utilize multiple received signals. Summary of the Invention

[0003] The purpose of this application is to provide an integrated array visible light communication receiver, receiving method and communication system, so as to expand the range of available incident positions and incident angles without increasing the photosensitive area of ​​a single high-speed detection unit, and to retain and utilize the channel state differences between multiple receiving branches formed by different light emission positions.

[0004] This application provides an integrated array visible light communication receiver based on large-area light collection and light field redistribution, comprising: A fluorescent antenna is disposed at the incident end of the receiver and has a light-receiving area and a waveguide light-out end face. The fluorescent antenna is used to receive first-band signal light incident at different positions and angles in the light-receiving area, convert the first-band signal light into a second-band fluorescent signal, and make the second-band fluorescent signal propagate in multiple directions in the planar waveguide structure of the fluorescent antenna and then output it from different positions of the waveguide light-out end face. An integrated silicon photomultiplier tube array detector, comprising arrays arranged along the waveguide end face. N r One detection unit, N r ≥ 2; Each of the detection units is optically coupled to different light-emitting positions of the waveguide light-emitting end face, and each independently converts the second band fluorescence signal coupled to itself into a receiving branch signal; Multi-channel analog front end, including with N r Each of the detection units is connected to N r Each analog signal processing channel amplifies the corresponding received branch signal and keeps each received branch signal independent of the others in the analog domain. A multi-channel synchronous acquisition module is used to synchronously sample the receiving branch signals output by the multi-channel analog front end based on the same sampling clock, so as to obtain multiple digitized receiving branch signals. The digital signal processing module is used for: synchronizing and gain normalizing each of the digitized receive branch signals; performing waveform-level equalization based on a neural network on each of the synchronized and gain-normalized receive branch signals before combining the multiple signals, to compensate for the nonlinear distortion and memory effect of each receive channel, and obtaining the compensated signals; performing discrete multi-tone demodulation on each of the compensated signals to obtain the demodulated symbols, wherein the discrete multi-tone demodulation includes serial-to-parallel conversion, removal of cyclic prefix, fast Fourier transform, downsampling, and zero-forcing equalization; estimating the channel quality of each receive channel using pilot signals; determining the combining weight of each receive channel based on the channel quality, and performing weighted combining of the demodulated symbols in the symbol domain to obtain the combined output; and performing parallel-to-serial conversion and demapping on the combined output to recover the original data signal from the transmitting end.

[0005] In a preferred embodiment, the fluorescent antenna includes an incident window, a fluorescence conversion layer, the planar waveguide structure, a filter layer, a reflective layer, and an encapsulation structure; The first band signal light enters the fluorescent antenna through the incident window, is absorbed by the fluorescent conversion layer and converted into a second band fluorescent signal, and the band of the second band fluorescent signal matches the response band of the detection unit; The second-band fluorescence signal propagates in multiple directions within the planar waveguide structure and couples from different positions on the waveguide end face to... N r The aforementioned detection unit.

[0006] In a preferred embodiment, the photosensitive area of ​​a single detection unit is smaller than the light-receiving area of ​​the light-receiving region; N r The detection units are arranged in parallel along the waveguide end face, and the number and arrangement of the detection units are determined according to the size of the waveguide end face. N r The total receiving width of each of the detection units is matched with the width of the waveguide end face, so that the second band fluorescence signals distributed and output from different positions of the waveguide end face are coupled to different detection units respectively.

[0007] In a preferred embodiment, the device further includes a microlens array disposed between the waveguide end face and the integrated silicon photomultiplier tube array detector, wherein each microlens in the microlens array is connected to... N r Each of the detection units corresponds to a specific microlens, and each microlens converges the second-band fluorescence signal output from the corresponding position of the wavelength-derived light-emitting end face to the photosensitive surface of the corresponding detection unit.

[0008] In a preferred embodiment, each of the analog signal processing channels further performs impedance matching and gain adjustment on the corresponding receiving branch signal; The integrated silicon photomultiplier tube array detector and the multi-channel analog front end are disposed on the same printed circuit board, and the microlens array is mounted on an optical bracket that is fixed relative to the printed circuit board. The microlens array, the integrated silicon photomultiplier tube array detector, and the multi-channel analog front end constitute a board-level integrated receiving component.

[0009] In a preferred embodiment, the waveform-level equalization is performed by a neural network equalizer based on a bidirectional gated cyclic unit. The neural network equalizer takes multiple consecutive received sample values ​​in a sliding window in the neighborhood of the sampling time to be equalized as input, extracts the forward time correlation features and backward time correlation features of the received waveform through the bidirectional gated cyclic unit, and outputs the compensated signal sample value corresponding to the sampling time to be equalized.

[0010] In a preferred embodiment, each of the detection units is composed of multiple micro-units, which recover their gain after an avalanche response through a recovery process with a recovery time constant. The nonlinear distortion and memory effect include at least: gain compression caused by the limited number of micro-units, in which the output amplitude deviates from a linear proportional relationship as the incident light power increases; and dynamic waveform distortion with time memory characteristics caused by the recovery process of the micro-units, in which the current output amplitude is affected by the previous input waveform.

[0011] In a preferred embodiment, let the first... i The demodulated symbol obtained after the received branch signal is processed by the waveform level equalization and the discrete multi-tone demodulation is: Tx i ( k , m The first, estimated using the pilot signal i The non-negative channel quality index for each receiving channel is: q i ,and If the combined weight is greater than 0, then the combined weight is... w i Determined according to the following formula: The digital signal processing module performs a weighted merging of the demodulated symbols in the symbol domain according to the following formula: in, i and j These are all sequence numbers of the receiving channel. i , j = 1, 2, ⋯, N r ; N r The number of the detection units; k Subcarrier sequence number; m For discrete polyphonic symbols; Tx i ( k , m ) is the first i The first branch signal obtained after the waveform level equalization and discrete multi-tone demodulation of the received branch signal is... m In the discrete polyphonic symbols, the first... k Demodulated symbols on each subcarrier; q i In order to utilize the pilot signal according to the first i The non-negative channel quality index is determined by the channel response amplitude or signal-to-noise ratio of each receiving channel after the waveform level equalization and the zero-forcing equalization.w i For the first i The merging weight of each receiving channel; Tx ( k , m ) is the first in the merged output m In the discrete polyphonic symbols, the first... k Combined symbols on each subcarrier.

[0012] This application provides an integrated array visible light communication receiving method based on large-area light collection and light field redistribution. The method is implemented based on the aforementioned receiver and includes: Step 100: The fluorescent antenna receives first-band signal light incident at different positions and angles in the light-receiving area, converts the first-band signal light into a second-band fluorescent signal, and causes the second-band fluorescent signal to propagate in multiple directions in the planar waveguide structure before being distributed and output from different positions on the waveguide light-emitting end face. Step 200: Each of the aforementioned detection units independently performs photoelectric conversion on the second band fluorescence signal coupled to itself obtained in step 100, and outputs... N r The path receives branch signals; Step 300: The multi-channel analog front end amplifies each of the received branch signals output in step 200, and the multi-channel synchronous acquisition module synchronously samples each of the received branch signals based on the same sampling clock to obtain multiple digitized received branch signals; Step 400: Synchronize and normalize the digitized receiving branch signals obtained in Step 300 respectively; before combining the multiple signals, perform waveform-level equalization based on neural networks on each of the synchronized and normalized receiving branch signals to compensate for the nonlinear distortion and memory effect of each receiving channel, and obtain the compensated signals; perform discrete multi-tone demodulation on each of the compensated signals to obtain the demodulated symbols, and use the pilot signal to estimate the channel quality of each receiving channel. The discrete multi-tone demodulation includes serial-to-parallel conversion, removal of cyclic prefix, fast Fourier transform, downsampling and zero-forcing equalization. Step 500: Determine the combining weight of each receiving channel based on the channel quality obtained in step 400, and perform weighted combining of each demodulated symbol obtained in step 400 in the symbol domain to obtain the combined output; Step 600: Perform parallel-to-serial conversion and demapping on the merged output obtained in step 500 to recover the original data signal from the transmitting end.

[0013] This application provides a visible light communication system, including a transmitter and the aforementioned receiver; The transmitting end is used to modulate the data to be transmitted onto the first band signal light in a discrete multi-tone modulation method, and to transmit the modulated first band signal light into free space. The receiver is used to receive the first band signal light that has been transmitted in free space and incident on the light-receiving area, and to recover the data to be transmitted.

[0014] The integrated array visible light communication receiver provided in this application uses a fluorescent antenna with a light-receiving area and a waveguide light-emitting end face as the receiving front end. It converts first-band signal light incident at different positions and angles into second-band fluorescent signals, distributing them from different positions on the waveguide light-emitting end face, allowing multiple detection units to share the same large light-receiving area. Based on this, each detection unit independently outputs a receiving branch signal, which is amplified through independent analog signal processing channels and synchronously sampled based on the same sampling clock to obtain multiple digital receiving signals with a consistent time base. This preserves the differences in power, delay, channel response, and distortion at different light-emitting positions before digital processing. The digital signal processing module processes signals in the following order: synchronization and gain normalization, then waveform-level equalization, followed by discrete multi-tone demodulation and channel quality estimation using pilots, and finally, channel quality-weighted combining in the symbol domain. The combining weights are determined after each receiving branch has completed nonlinearity and memory effect compensation. Therefore, without increasing the photosensitive area of ​​a single high-speed detection unit or reducing its modulation bandwidth, it helps to expand the range of available incident positions and incident angles, improve the uniformity of data transmission rate and reception stability under conditions of spot position shift or incident angle change, and reduce the difficulty of beam alignment between the transmitter and receiver.

[0015] In a fluorescent antenna scheme comprising an incident window, a fluorescence conversion layer, a planar waveguide structure, a filter layer, a reflective layer, and an encapsulation structure, the band of the second-band fluorescence signal matches the response band of the detection unit, which helps to increase the proportion of the converted fluorescence signal effectively received by the detection unit. The filter layer reduces the amount of unconverted first-band signal light and other non-target band light entering the detection array, and the reflective layer guides some of the fluorescence propagating in the non-emitting direction back to the planar waveguide structure. The combination of these two elements helps to improve fluorescence utilization and reduce the adverse effects of background light.

[0016] In a scheme where the photosensitive area of ​​a single detector unit is smaller than the light-receiving area of ​​the light-receiving region, and the total receiving width of multiple detector units matches the width of the waveguide light-exit end face, the second-band fluorescence signals distributed from different positions on the waveguide light-exit end face can be coupled to different detector units respectively, rather than being concentrated and coupled to one detector unit. This forms multiple spatially separated receiver sub-apertures, providing a structural basis for subsequent multi-branch joint processing.

[0017] Setting up microlens arrays corresponding to each detection unit helps to converge the second-band fluorescence signal output from the corresponding position of the waveguide end face into the photosensitive surface of the corresponding detection unit, reducing coupling loss. Placing the integrated silicon photomultiplier tube array detector and the multi-channel analog front end on the same printed circuit board and fixing them relative to each other with optical brackets and microlens arrays to form a board-level integrated receiving component helps to shorten the signal path between each detection unit and the corresponding analog front end, maintaining the consistency of response and assembly stability of each receiving channel. Each analog signal processing channel performs impedance matching and gain adjustment on the corresponding receiving branch signal, which makes each signal within a comparable amplitude range, providing conditions for subsequent channel quality comparison and weighted combining.

[0018] A neural network equalizer based on a bidirectional gated cyclic unit is employed. Taking multiple consecutive received sample values ​​within a sliding window in the neighborhood of the sampling time to be equalized as input, the forward and backward time correlation features of the received waveform are extracted. This compensates for dynamic waveform distortion with time memory characteristics, where the current output is affected by the previous input waveform. This compensation process corresponds to the device mechanism of power-related gain compression caused by the limited number of silicon photomultiplier tube microcells and memory waveform distortion caused by the microcell recovery process. It is specifically targeted and helps to reduce gain compression and memory effects in each receiving branch, making the equivalent response of each receiving branch closer to the linear condition that facilitates weight allocation according to channel quality.

[0019] Normalization is performed using non-negative channel quality metrics to determine the combining weights, ensuring that each combining weight is non-negative and their sum is one. This provides a unified normalization benchmark for the contribution relationships of all branches, with receiving branches with higher channel quality contributing more to the combined output and receiving branches with lower channel quality contributing less. Simultaneously, weighted combining is applied to the demodulated symbols obtained after waveform-level equalization and discrete multi-tone demodulation, rather than the undemodulated time-domain sampling points. This ensures that channel quality evaluation and combining weight determination are based on the completion of compensation and zero-forcing equalization in each receiving branch, helping to reduce the impact of receiver link distortion on weight allocation and improving the reliability of the combined output.

[0020] The corresponding receiving method processes the physically received signal sequentially according to the above order, forming a complete signal processing chain from light collection, wavelength conversion, independent photoelectric conversion, independent analog processing, synchronous sampling, to per-channel compensation, per-channel demodulation, channel quality estimation, symbol domain weighted combining, and demapping. Its processing object is the electrical signal with a clear physical meaning generated by the optical communication receiving link. The corresponding visible light communication system includes a transmitter and the aforementioned receiver, forming a complete link between the transmitter and receiver using discrete multi-tone modulation. This helps maintain the effectiveness of the communication link when the incident position, incident angle, transmission distance, and ambient light conditions change, making it suitable for visible light communication applications with high requirements for beam alignment and receiving field of view.

[0021] In summary, this application, through the synergistic cooperation of the aforementioned technical means, helps to alleviate the mutual constraints between transmission rate, effective receiving area, and receiving field of view in high-speed visible light reception, and improves the availability of the receiver under actual deployment conditions.

[0022] The specification of this application contains numerous technical features distributed across various technical solutions. Listing all possible combinations of these technical features (i.e., technical solutions) would make the specification excessively lengthy. To avoid this problem, the various technical features disclosed in the above-described invention, the various technical features disclosed in the following embodiments and examples, and the various technical features disclosed in the accompanying drawings can be freely combined to form various new technical solutions (all of which are considered to have been described in this specification), unless such a combination of technical features is technically infeasible. For example, one example discloses feature A+B+C, and another example discloses feature A+B+D+E. Features C and D are equivalent technical means that serve the same function, and technically only one needs to be used; they cannot be used simultaneously. Feature E can technically be combined with feature C. Therefore, the solution A+B+C+D should not be considered as described because it is technically infeasible, while the solution A+B+C+E should be considered as described. Attached Figure Description

[0023] Figure 1 is a schematic diagram of the structure of an integrated array visible light communication receiver based on large-area light collection and light field redistribution provided in one embodiment of this application.

[0024] Figure 2 is a flowchart illustrating an integrated array visible light communication receiving method based on large-area light collection and light field redistribution provided in one embodiment of this application.

[0025] Figure 3 is a schematic diagram of the three-dimensional structure of the receiver and the relationship between the multi-branch receiving signal processing provided in one embodiment of this application, wherein a is a schematic diagram of the three-dimensional structure of the fluorescent antenna, microlens array and integrated silicon photomultiplier tube array detector, and b is a schematic diagram of the signal processing relationship of the multi-path receiving branches.

[0026] Figure 4 is a schematic diagram of a test system for a visible light communication system provided in one embodiment of this application, wherein a is a schematic diagram of the signal processing flow and optical path connection between the transmitter and receiver, b is a schematic diagram of the overall arrangement of the test optical path, and c is a schematic diagram of the arrangement of the board-level integrated receiver component.

[0027] Figure 5 is a schematic diagram of the receiving link response when the incident light spot position changes in one embodiment of this application, wherein a is a schematic diagram of the frequency response of the receiving link under different incident light spot displacements, and b is a schematic diagram of the normalized time-domain impulse response of the receiving link under different incident light spot displacements.

[0028] Figure 6 is a schematic diagram showing the distribution of achievable data transmission rates with incident spot position under different incident spot sizes in one embodiment of this application. In the figure, a to c are schematic diagrams of the rate distribution corresponding to the multi-receiver branch weighted merging mode, the single center detection unit mode, and the single edge detection unit mode under the first spot size, respectively. d to f are schematic diagrams of the rate distribution corresponding to the above three modes under the second spot size, respectively. g to i are schematic diagrams of the rate distribution corresponding to the above three modes under the third spot size.

[0029] Figure 7 is a schematic diagram of the communication performance of a visible light communication system provided in an embodiment of this application under different test conditions. In this diagram, a is a schematic diagram showing the achievable data transmission rate as a function of transmission distance under different incident light spot sizes; b is a schematic diagram showing the modulation order distribution and signal-to-noise ratio distribution of each effective subcarrier under different ambient light intensities; c is a schematic diagram showing the bit error rate as a function of transmission distance with and without waveform level equalization under different incident light spot sizes; d is a schematic diagram showing the achievable data transmission rate of four processing configurations under different ambient light intensities; e is a schematic diagram showing the achievable data transmission rate and bit error rate in repeated testing; f is a schematic diagram showing the transmission rate reduction as a function of ambient light in two receiving methods: with and without a fluorescent antenna; g is a schematic diagram showing the achievable data transmission rate as a function of incident angle for different incident light spot sizes in single-receiver branch mode; and h is a schematic diagram showing the achievable data transmission rate as a function of incident angle for different incident light spot sizes in multi-receiver branch weighted merging mode. Detailed Implementation

[0030] In the following description, many technical details are presented to help the reader better understand this application. However, those skilled in the art will understand that the technical solutions claimed in this application can be implemented even without these technical details and various variations and modifications based on the following embodiments.

[0031] Explanation of some concepts Large-area light collection and light field redistribution refers to an optical processing method that uses a light-receiving area larger than the photosensitive area of ​​a single detection unit to receive incident communication light, and reorganizes the propagation path and spatial energy distribution of the incident communication light so that signal light incident on different positions of the light-receiving area and propagating in different directions can be distributed and output from different positions of a predetermined light-output area and coupled to multiple detection units respectively.

[0032] A fluorescent antenna is an optical front-end device that includes a fluorescent conversion layer and a planar waveguide structure. It can convert a first-band signal light into a second-band fluorescent signal and distribute the fluorescent signal at different positions on the waveguide end face after propagation through the planar waveguide. It is a typical optical front-end for realizing large-area light collection and light field redistribution.

[0033] The light-receiving area refers to the region in a fluorescent antenna used to receive first-band signal light incident at different positions and angles, and its light-receiving area is larger than the photosensitive area of ​​a single detection unit.

[0034] The waveguide end face refers to the end face of a fluorescent antenna from which the second-band fluorescent signal propagating within the planar waveguide structure is distributed and output from different positions, and optically coupled with multiple detection units.

[0035] The first-band signal light refers to the optical signal emitted by the visible light communication transmitter that carries the data to be transmitted.

[0036] The second-band fluorescence signal refers to the fluorescence signal obtained after the first-band signal light is absorbed and converted by the fluorescence conversion layer.

[0037] The term "band matching" means that at least a portion of the emission band of the second band fluorescence signal is located within the effective response band of the detection unit, so that the second band fluorescence signal can be received by the detection unit.

[0038] The receiving branch signal refers to an electrical signal that is independently output by a single detection unit after photoelectric conversion of the second-band fluorescence signal coupled to itself, and is independently transmitted and processed along the corresponding analog signal processing channel.

[0039] The receiving channel refers to the signal processing path consisting of a detection unit, an analog signal processing channel corresponding to the detection unit, a synchronous acquisition path, and a digital processing branch.

[0040] A receiver sub-aperture unit refers to a single detection unit that retains an independent electrical signal output and is optically coupled to different light-emitting positions on the waveguide end face; multiple receiver sub-aperture units together constitute an arrayed multi-branch receiver structure.

[0041] Synthetic aperture reception refers to a reception method in which the electrical signals output by multiple spatially separated receiving sub-aperture units are jointly processed in the digital domain under a direct detection system to obtain an effective reception capability greater than the combined output of a single detection unit. The synthetic aperture reception does not require the recovery of the original optical carrier phase, nor does it require coherent superposition of the optical wavefront as a necessary condition.

[0042] Waveform-level equalization refers to the equalization process performed on each receiving branch before combining multiple signals, using the continuous sampled values ​​of the receiving branch as the processing object. This process aims to reduce power-related gain compression and dynamic waveform distortion with time memory characteristics.

[0043] Discrete multi-tone demodulation refers to the process of sequentially performing serial-to-parallel conversion, removing cyclic prefixes, performing fast Fourier transform, downsampling, and zero-forcing equalization on the compensated signal to obtain the corresponding discrete multi-tone symbols and demodulated symbols on the subcarriers. When the transmitter adopts a Hermitian symmetric structure, the downsampling refers to extracting the non-redundant effective subcarriers corresponding to the Hermitian symmetric structure after the fast Fourier transform.

[0044] Channel quality metrics are non-negative indicators used to characterize the reception status of a receiving channel, determined using pilot signals and based on the channel response amplitude, channel response power, or signal-to-noise ratio after waveform-level equalization and zero-forcing equalization of the corresponding receiving channel. q i All receiving channels use the same index definitions and calculation scales.

[0045] Symbol-domain weighted combining refers to determining the combining weights based on the channel quality of each receiving channel after waveform-level equalization and discrete multi-tone demodulation have been completed in each receiving branch. w i And for the demodulated symbols corresponding to the same discrete multitone symbols and the same subcarriers Tx i ( k , m Perform a weighted summation to obtain the merged symbol. Tx ( k , m ) processing.

[0046] The negative 3 dB receiving field of view (NDF) refers to the angular range between the positive and negative incident angles when, under predetermined test conditions, the achievable data transmission rate drops to approximately half of its maximum value within the incident angle scanning range, using the maximum achievable data transmission rate as a reference. The negative 3 dB half-field of view (FFB) refers to the absolute value of the corresponding negative 3 dB boundary angle.

[0047] The rate-area product is the product of the average achievable data transmission rate at each measuring point within a predetermined performance evaluation area and the area of ​​that performance evaluation area.

[0048] Matching the total receiving width with the width of the waveguide light-emitting end face means that the coverage of the photosensitive area of ​​multiple detection units or the effective aperture of their corresponding microlenses in the width direction of the waveguide light-emitting end face can cover the main light-emitting area of ​​the waveguide light-emitting end face, so that the second-band fluorescence signal output from different light-emitting positions can be coupled to at least one detection unit respectively.

[0049] Technical concept In summary, the technical concept of this application lies in constructing a sequential processing link for receiving scenarios characterized by "same light-receiving area, multiple light-emitting positions, and inconsistent branch states" formed at the front end of large-area light collection and light field redistribution. The fluorescent antenna 10, through the light-receiving area 101, the fluorescent conversion layer 12, and the planar waveguide structure 13, converts first-band signal light incident at different positions and angles into second-band fluorescent signals, distributing them out from different positions on the waveguide light-emitting end face 102; thereby, the receiving branch signals output by each detection unit 21 coupled to different light-emitting positions... y i ( n The power, delay, response, and distortion vary depending on the propagation path and coupling state. This application does not directly parallel the outputs of each detection unit 21 in the analog domain. Instead, these differences are prevented from being fixedly superimposed before entering the digital signal processing module 60, as they are transmitted through independent analog signal processing channels 41 in the multi-channel analog front-end 40 and a multi-channel synchronous acquisition module 50 based on the same sampling clock. The gain compression caused by the limited number of micro-units in the silicon photomultiplier tube and the memory waveform distortion caused by the micro-unit recovery process will make the linearity and noise assumptions relied upon by classical maximum ratio combining not fully valid in the actual operating range of the device. Therefore, this application first normalizes the synchronization and gain of each signal to... ȳ i ( n Before merging multiple paths, waveform-level equalization based on neural networks is performed path by path to obtain... ỹ i ( n Subsequently, discrete multi-tone demodulation was performed path by path to obtain the demodulated symbols. Txi ( k , m The channel quality index is estimated after compensation using pilot signals. q i Determine the merge weights w i Ultimately, in the symbol field, for each path Tx i ( k , m Weighted merging yields Tx ( k , m This "first-path compensation, then post-compensation quality weighting" arrangement allows the channel quality upon which the weight allocation is based to better reflect the actual contribution of each receiving branch to the recovered data. Comparative tests in Example 3 also show that, under the corresponding test conditions, the achievable data transmission rate obtained by jointly employing path-by-path waveform level equalization and post-compensation quality weighting is higher than the results obtained by employing only one of these processing measures. Therefore, front-end optical field redistribution, independent branch retention, pre-merging path-by-path compensation, and symbol domain quality weighting merging form an interconnected overall processing link around the concept of "preserving differences, correcting differences, evaluating and utilizing differences," which helps to expand the effective receiving area and incident angle adaptation range without increasing the photosensitive area of ​​a single high-speed detection unit.

[0050] Furthermore, through long-term and in-depth research, the inventors of this application discovered that the aforementioned problems do not stem from the insufficient performance of a single device, but are closely related to the overall design of the receiving structure and signal processing method. Traditional receiving structures concentrate light collection and detection on the same small-area device and rely on a single optical axis focusing, so that improving gain often comes at the cost of stringent alignment requirements. The inventors realized that if the incident light is first collected by a large-area optical front end, and its propagation path and spatial energy distribution are reorganized into a light field distributed from different positions on a predetermined light-emitting end face, the light collection link and the high-speed detection link can be decoupled. The key to the problem is not simply increasing the number of detection units, but retaining and utilizing the differences between the branches after expanding the receiving range. The inventors also found that if the outputs of multiple detection units are directly connected in parallel in the analog domain, or mixed before channel quality evaluation, the effective signals and noise of each branch will be fixedly superimposed, and the differences in power, delay, response, and distortion corresponding to different light-emitting positions that could have been utilized will be discarded. This is precisely why existing multi-detector schemes cannot fully benefit from the differences. More importantly, after analyzing the mechanism of silicon photomultiplier tube devices, the inventors realized that the limited number of micro-units causes power-related gain compression, and the recovery process causes memory waveform distortion. These two factors mean that the applicability of channel quality estimation based on linear assumptions and classical maximum ratio merging is not entirely valid in the actual operating region of the device. Furthermore, if the quality of each branch is estimated and weighted without compensation, the distortion will also affect the rationality of the weight allocation. Based on the above in-depth research, the inventors of this application propose a technical solution that uses a large-area light collection and light field redistribution optical front end as a shared receiving front end, retains multiple independent receiving branches, performs waveform-level equalization based on neural networks for each branch before merging, and performs weighted merging in the symbol domain according to channel quality after compensation. The implementation process of this application is described in detail below through specific embodiments.

[0051] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0052] The technical solution of this application will be further described below with reference to embodiments. It should be understood that the following embodiments are used to illustrate the technical concept and its possible implementation methods of this application, and are not intended to unnecessarily limit the scope of protection of this application. Where there is no technical conflict, the technical features in the different embodiments below can be combined with each other.

[0053] Example 1 As shown in Figure 1, this embodiment provides an integrated array visible light communication receiver based on large-area light collection and light field redistribution. The receiver includes a fluorescent antenna 10, an integrated silicon photomultiplier tube array detector 20, a multi-channel analog front-end 40, a multi-channel synchronous acquisition module 50, and a digital signal processing module 60. Optionally, the receiver further includes a microlens array 30 disposed between the fluorescent antenna 10 and the integrated silicon photomultiplier tube array detector 20. The three-dimensional structural relationship of the fluorescent antenna 10, the microlens array 30, and the integrated silicon photomultiplier tube array detector 20, as well as the signal processing relationship of the multiple receiving branches, can also be seen in Figure 3.

[0054] In this application, "large-area light collection and light field redistribution" refers to: using a light-receiving area larger than the photosensitive area of ​​a single detector unit to receive incident communication light, and reorganizing the propagation path and spatial energy distribution of the incident communication light, so that signal light incident at different positions and propagating in different directions within the light-receiving area can be distributed and output from a predetermined light-emitting area, and coupled to multiple detector units respectively. Thus, multiple detector units can share the same large-area light-receiving area without the need for separate large-area focusing optical systems.

[0055] In this embodiment, a fluorescent antenna 10 is used as the optical front end for large-area light collection and light field redistribution. The fluorescent antenna 10 is disposed at the incident end of the receiver and has a light-receiving area 101 and a waveguide light-emitting end face 102. Specifically, the fluorescent antenna 10 includes an incident window 11, a fluorescence conversion layer 12, a planar waveguide structure 13, a filter layer 14, a reflective layer 15, and an encapsulation structure 16.

[0056] First-band signal light incident at different positions and angles within the light-receiving area enters the fluorescent antenna through the incident window and is absorbed and converted into a second-band fluorescent signal by the fluorescent conversion layer. The band of the second-band fluorescent signal matches the response band of the downstream silicon photomultiplier tube detection unit, ensuring that the converted fluorescent signal can be effectively received by the corresponding detection unit. For example, the first-band signal light can be short-wavelength visible light output from a visible light communication transmitter, and the second-band fluorescent signal can be a fluorescent signal with a peak wavelength in the green visible light band.

[0057] The converted second-band fluorescence signal propagates in multiple directions within the planar waveguide structure and reaches different positions on the waveguide end face under the combined effects of total internal reflection, reflection, and propagation within the fluorescence material. The filter layer reduces the amount of first-band signal light or other non-target band light that is not absorbed by the fluorescence conversion layer from entering the detector array. The reflective layer reflects a portion of the fluorescence propagating in the non-emitting direction back into the planar waveguide structure. The encapsulation structure maintains the relative positions of the fluorescence conversion layer, the planar waveguide structure, the filter layer, and the reflective layer and provides mechanical protection. The location of the filter layer can be varied. In one optional embodiment, the filter layer is positioned in the optical path of the second-band fluorescence signal emanating from the integrated silicon photomultiplier tube array detector (see Figure 1) to filter out the first-band signal light and other non-target band light that are not absorbed by the fluorescence conversion layer before the fluorescence signal enters the detector array. In another optional embodiment, the filter layer is disposed on the light incident side, in the optical path between the incident window and the fluorescence conversion layer, to filter out background light outside the target receiving band before the incident light reaches the fluorescence conversion layer; in this method, the filter layer and the fluorescence conversion layer are disposed sequentially along the incident direction, and the reflective layer is disposed on the side of the fluorescence conversion layer opposite to the incident direction. Either of the above two configuration methods can be used, or a combination thereof.

[0058] It should be noted that this embodiment does not rely on directly focusing the incident light onto a single small-area detection unit along a single optical axis. Instead, it first collects the incident light through a larger light-receiving area, and then uses fluorescence conversion and planar waveguide propagation to distribute the signal light out at different positions on the waveguide end face. Therefore, when the position of the incident light spot, the incident angle, or the receiver orientation changes, the incident communication light can still reach one or more detection units through fluorescence conversion and planar waveguide propagation, thereby expanding the range of incident positions and incident angles that the receiver can utilize.

[0059] The integrated silicon photomultiplier tube array detector 20 includes arrays arranged along the waveguide end face 102. N r There are 21 detection units, among which... N r The number of detection units 21, and N r ≥ 2. This embodiment exemplarily adopts a row of multiple detector units, with each detector unit arranged in parallel along the extension direction of the waveguide light-emitting end face and optically coupled to different light-emitting positions of the waveguide light-emitting end face.

[0060] The photosensitive area of ​​a single detector element is smaller than the light-receiving area of ​​the fluorescent antenna. The number and arrangement of multiple detector elements can be determined based on the width of the waveguide end face, the width of the photosensitive surface of a single detector element, and the installation interval between adjacent detector elements, so that the total receiving width formed by multiple detector elements matches the width of the waveguide end face. Therefore, the second-band fluorescence signal output from different positions on the waveguide end face can be coupled to different detector elements respectively, rather than being concentrated and coupled to only one detector element.

[0061] Each detection unit independently converts the second-band fluorescence signal coupled to it into an electrical signal, forming a corresponding receiving branch signal. Because different detection units correspond to different positions on the wavelet output facets, the propagation path, propagation distance, optical loss, and coupling state experienced by the second-band fluorescence signal when reaching different detection units may differ. Therefore, the received power, bandwidth, time delay, noise level, and nonlinear distortion of each receiving branch signal may also differ. Furthermore, the relative strength relationship between the receiving branches may change when the incident spot position or incident angle changes.

[0062] This embodiment does not directly connect the outputs of multiple detection units in parallel in the analog domain. Instead, it retains the independent output of each detection unit, allowing each detection unit to form a separate receiving sub-aperture unit. Multiple receiving sub-aperture units together constitute an arrayed multi-branch receiving structure. The term "synthetic aperture reception" as used herein refers to the digital domain joint processing of the electrical signals output from multiple spatially separated receiving sub-aperture units under a direct detection system to obtain an effective receiving capability greater than the combined output of a single detection unit. Synthetic aperture reception does not require the recovery of the original optical carrier phase, nor does it require coherent superposition of the optical wavefront as a necessary condition.

[0063] Furthermore, the microlens array 30 is disposed between the waveguide end face 102 of the fluorescent antenna 10 and the integrated silicon photomultiplier tube array detector 20. Each microlens 31 in the microlens array 30 corresponds one-to-one with each detection unit 21. Each microlens converges the second-band fluorescence signal output from the corresponding position of the waveguide end face to the photosensitive surface of the corresponding detection unit, thereby reducing coupling loss caused by mismatch in size or spacing between the waveguide end face and the photosensitive surface of the detection unit.

[0064] The integrated silicon photomultiplier tube array detector 20 and the multi-channel analog front-end 40 are mounted on the same printed circuit board 70, and the microlens array 30 is mounted on an optical support 80 that is fixed relative to the printed circuit board 70. The optical support defines the relative distance and lateral position between the microlens array and the silicon photomultiplier tube array, so that the emitted light spots of each microlens fall into the photosensitive surface of the corresponding detection unit. Thus, the microlens array, the integrated silicon photomultiplier tube array detector, and the multi-channel analog front-end together constitute a board-level integrated receiving component.

[0065] The multi-channel analog front-end 40 includes and N r Each detection unit 21 is connected to N r Each analog signal processing channel 41 amplifies, impedance-matches, and adjusts the gain of its corresponding received branch signal, ensuring that each received branch signal remains independent in the analog domain and is not mixed before channel quality evaluation. Each analog signal processing channel can employ the same nominal circuit structure, or channel calibration can be performed based on device differences between the detection units to ensure that each received branch signal is within an amplitude range suitable for subsequent synchronous sampling and digital processing.

[0066] Compared to methods where multiple detector units are directly electrically connected in parallel to the same analog front-end, the independent analog signal processing channel in this embodiment does not uniformly mix the effective signals, background light noise, dark count noise, and circuit noise of each detector unit before channel quality evaluation. Furthermore, the output capacitors of each detector unit are not all directly and centrally loaded onto the same analog input node. These differences enable the digital signal processing module to perform synchronization, gain normalization, waveform-level equalization, and channel quality estimation separately for each receiving branch.

[0067] The multi-channel synchronous acquisition module 50 synchronously samples the received branch signals output by the multi-channel analog front-end 40 based on the same sampling clock, obtaining multiple digitized received branch signals. By using the same sampling clock, the relative time drift caused by different independent sampling time bases of different received branches can be reduced, providing a time reference for subsequent synchronous processing of each signal, channel quality comparison, and symbol domain weighted merging.

[0068] In the experimental system built in this embodiment, the multi-channel synchronous acquisition module 50 is implemented using a multi-channel oscilloscope. The oscilloscope synchronously acquires the voltage waveforms of each receiving branch and inputs the sampled data into a back-end computer, which then performs the following digital signal processing offline. This offline processing method can be used to verify the processing flow and its technical effects. In practical applications, the digital signal processing module 60 can also be implemented by a processor executing program instructions in memory, or by a field-programmable gate array, a digital signal processor, or an application-specific integrated circuit.

[0069] For any i i The equivalent reception relationship before and after synchronous sampling for each receiving branch can be expressed as: in, i To receive the branch sequence number, i = 1, 2, ⋯, N r ; n The sequence number of the discrete sampling time; x ( n () is the reference signal corresponding to the time-domain signal of the transmitting end; y i ( n ) is the first i The receiving branch signal is digitized; g i (·) is the first i The equivalent response of the receiving link includes the amplitude response, frequency response, and nonlinear response formed by fluorescence conversion, planar waveguide propagation, microlens coupling, silicon photomultiplier tube photoelectric conversion, and analog front-end processing. z i ( n ) is the first i Noise components in the receiving branch of the path.

[0070] Silicon photomultiplier tubes (SMTs) consist of a finite number of microcells. As the number of photons incident on the detector unit increases per unit time, the number of triggered microcells may gradually approach the total number of available microcells, causing a decrease in the increment of the output amplitude relative to the incident light power, manifested as power-dependent gain compression. On the other hand, after an avalanche response, the microcells need to undergo a recovery process characterized by a recovery time constant to recover the corresponding gain. When the time interval between adjacent signal changes is close to the time scale of the microcell recovery process, the current output may also be affected by the previous input waveform, manifesting as dynamic waveform distortion with time memory characteristics such as waveform tailing, amplitude history dependence, or intersymbol interference.

[0071] Classical maximum ratio combining is typically based on the assumption that the responses of each branch are approximately linear and that channel quality can be reliably estimated. However, in the actual operating range of silicon photomultiplier tubes (SMTs), these linearity and noise assumptions may not hold entirely. If the channel quality of each receiver branch is directly evaluated and weighted combining is performed without path-by-path compensation, power-related gain compression and memory effects may affect the channel quality estimation results, and receiver branches with different levels of distortion may interfere with each other after combining.

[0072] Therefore, in this embodiment, the digital signal processing module 60 first performs synchronization and gain normalization on each digitized receiving branch signal, and then performs waveform-level equalization based on a neural network on each receiving branch signal before merging the multiple signals. The synchronization can utilize a known training sequence or pilot sequence inserted at the transmitter, and determine the starting position of the data block or discrete polyphonic symbol by calculating the correlation between the receiving branch signal and the known sequence; the gain normalization can determine a non-zero scaling factor based on the amplitude, effective value, or average power of the corresponding receiving branch in the training segment or pilot segment, and adjust the signal amplitude of the receiving branch using the scaling factor.

[0073] Let the first i The signal after synchronization and gain normalization of the receiving branch is: ȳ i ( n Waveform-level equalization based on neural networks can be expressed as: in, ỹ i ( n ) is the first i The receiving branch of the path is at the sampling time n The corresponding compensated signal sample value; F θi (·) is used for the first i Neural network waveform compensation mapping for the receiving branch; θ i These are the parameters of the corresponding neural network; W 1 and W 2 represents the number of sampling points in the input sliding window before and after the sampling time to be equalized, respectively. By moving the sliding window along the time axis of the received waveform, the compensated signal sample values ​​corresponding to each sampling time can be obtained sequentially.

[0074] In this embodiment, the neural network equalizer employs a network structure based on bidirectional gated recurrent units. The bidirectional gated recurrent units extract time-related features before and after the sampling time to be equalized, enabling the network to utilize not only the current sample value but also consecutive received sample values ​​before and after the current sample value to compensate for dynamic waveform distortion with time memory characteristics. The network output is a single compensated signal sample value corresponding to the sampling time to be equalized, and multiple compensated signal samples are arranged in chronological order to form a compensated continuous received waveform.

[0075] In one optional embodiment, the neural network equalizer includes an input layer, at least one bidirectional gated recurrent unit layer, and an output layer. The input layer receives continuously received sampled values ​​in the sliding window; the bidirectional gated recurrent unit layer extracts time-related features along both the forward and reverse directions; the output layer outputs a single compensated signal sample value corresponding to the sampling time to be equalized based on the time-related features. The length of the sliding window, the number of bidirectional gated recurrent unit layers, and the hidden state dimension can be determined based on the sampling rate, the time scale of the silicon photomultiplier tube microcell recovery process, and the equalization effect on the verification data.

[0076] In one optional training method, the training data includes receive branch waveforms acquired under different incident positions, incident angles, received power, or ambient light conditions, and the supervision label is a known training waveform from the transmitter that is time-aligned with the receive branch waveforms. Let a training batch include... B The nth sample, the th b The network output for each sample is sh b The corresponding supervision label is s b Then the mean squared error loss function can be used as shown in the following formula: in, L ( θ i ) represents network parameters θ i The corresponding loss function value, B This represents the number of samples in the training batch. b This refers to the sample sequence number. Gradient-based optimization algorithms can be used to update network parameters, and the training stopping condition can be determined based on the loss changes in the validation data. After training, the network parameters are fixed and used to equalize the sampled waveforms of the corresponding receiving branches. Each receiving branch can use its own corresponding network parameters; alternatively, a set of network parameters can be shared when each receiving channel is calibrated and the signal characteristics meet the sharing conditions.

[0077] The waveform-level equalization is not described as completely restoring the receive branches to an ideal linear channel, but rather as reducing power-related gain compression and memory effects, making the equivalent responses of each receive branch closer to the linear conditions that facilitate discrete multi-tone demodulation, pilot channel quality estimation, and subsequent weight allocation. By processing each receive branch separately before merging, the nonlinear distortions in different branches can be avoided by pre-fixing and superimposing them.

[0078] After waveform-level equalization, the digital signal processing module performs discrete multi-tone demodulation on each compensated signal. The discrete multi-tone demodulation sequentially includes serial-to-parallel conversion, cyclic prefix removal, fast Fourier transform, downsampling, and zero-forcing equalization. In this embodiment, the transmitter employs a Hermitian symmetric structure, and the downsampling refers to extracting the non-redundant effective subcarriers corresponding to the Hermitian symmetric structure after the fast Fourier transform. After the above processing, the demodulated symbols for each receiver branch on the corresponding discrete multi-tone symbols and subcarriers are obtained.

[0079] No. i The demodulated symbol of the receiving branch is denoted as Tx i ( k , m ),in, k Subcarrier sequence number, m This refers to the discrete multi-tone symbol number. The digital signal processing module 60 uses the known pilot signal inserted at the transmitter to estimate the channel response amplitude, channel response power, or signal-to-noise ratio of each receiving branch after waveform-level equalization and discrete multi-tone demodulation, and accordingly obtains the non-negative channel quality index for each receiving channel. q i For example, the average channel response power or average signal-to-noise ratio on multiple pilot subcarriers within a predetermined data block can be used as... q i The same index definition and calculation scale are applied to all receiving channels.

[0080] When the sum of the channel quality metrics of all receiving channels is greater than zero, the first... i The combined weight of each receiving channel w i Determined according to the following formula: in, q i For the first i Non-negative channel quality metrics for each receiving channel; j For the summation sequence number; N r Number of receive channels; w i For the first iThe combining weights of each receiving channel are determined according to the above formula. All combining weights are non-negative, and the sum of all combining weights is 1. Therefore, the receiving branch with higher channel quality contributes more to the combined output, while the receiving branch with lower channel quality contributes less.

[0081] The digital signal processing module 60 performs weighted merging of the demodulated symbols in the symbol field to obtain the merged symbol: in, Tx i ( k , m ) is the first i The receiving branch of the road is in the first m In the discrete polyphonic symbols, the first... k Demodulated symbols on each subcarrier; w i For the first i The merging weight of each receiving channel; Tx ( k , m ) is the first m In the discrete polyphonic symbols, the first... k Combined symbols on each subcarrier.

[0082] It is important to note that the aforementioned weighted combining applies to the symbol symbols obtained after waveform-level equalization and discrete multi-tone demodulation by each receiving branch, rather than directly to the undemodulated time-domain sampling points. This processing order ensures that channel quality evaluation and the determination of combining weights are based on the completion of nonlinearity and memory effect compensation and zero-forcing equalization by each receiving branch, which helps reduce the impact of receive link distortion on weight allocation. Before weighted combining, each receiving branch can use training sequences or pilot signals to align the data block start position, frequency offset, phase, and subcarrier index, ensuring that the demodulated symbols participating in weighted combining correspond to the same discrete multi-tone symbol, the same subcarrier, and the same constellation reference.

[0083] In one alternative implementation, channel quality indicators q i The weighting is determined based on the overall pilot response or overall signal-to-noise ratio of each receiving channel, so that the same combining weight is applied to the corresponding receiving channels within a predetermined data block or predetermined sampling record. In another optional implementation, the weighting is determined based on the pilot channel response on different subcarriers. i The receiving channel is in the first k Non-negative channel quality metrics on each subcarrier q i ( k The corresponding merging weights are determined according to the following formula: In this method, the above symbolic field weighted merging formula is used. w i Replace with w i ( k This allows for weighted combining based on channel differences across different subcarriers. Whether to use a unified weight for the data block or a per-subcarrier weight depends on the degree of channel difference between different subcarriers within the system bandwidth.

[0084] Optionally, when the channel quality of a certain receiving branch is lower than a predetermined threshold, shows obvious saturation, or cannot achieve reliable synchronization, the weight of that receiving branch can be reduced, or the receiving branch can be temporarily not used in this merging process. When the sum of the channel quality indicators of all receiving branches is zero or lower than a predetermined effective threshold, a no-signal flag can be output to avoid the denominator in the above merging weight calculation formula being zero. The above protection processing can be used as a boundary processing method in a practical receiving system.

[0085] After symbol-domain weighted combining, the digital signal processing module performs parallel-to-serial conversion and constellation demapping (hereinafter referred to as demapping) on ​​the combined output to recover the original data signal from the transmitter. Optionally, the bit error rate can be calculated based on the recovered data signal and a known test sequence, or the achievable data transmission rate can be estimated based on the signal-to-noise ratio and modulation order of each subcarrier, in order to evaluate the communication performance of the receiver under different incident positions, incident angles, ambient light levels, and transmission distances.

[0086] Example 2 As shown in Figure 2, this embodiment provides an integrated array visible light communication receiving method based on large-area light collection and light field redistribution. The method is implemented based on the receiver described in Embodiment 1 and includes steps 100 to 600.

[0087] Step 100: The fluorescent antenna receives first-band signal light incident at different positions and angles in the light-receiving area, converts the first-band signal light into a second-band fluorescent signal, and then distributes and outputs the second-band fluorescent signal at different positions on the waveguide end face of the fluorescent antenna after it propagates in multiple directions in the planar waveguide structure of the fluorescent antenna.

[0088] In step 100, the first-band signal light carries the data to be transmitted at the transmitting end. After entering the fluorescent antenna, the first-band signal light is absorbed by the fluorescent conversion layer and converted into a second-band fluorescent signal that matches the response band of the silicon photomultiplier tube detection unit. The second-band fluorescent signal is not limited to continuing to propagate along the incident direction, but reaches different positions on the waveguide end face of the waveguide through different propagation paths in the planar waveguide structure, thereby reorganizing the signal light incident at different positions and angles in a large light-receiving area into a distributed output light field that can cover multiple detection units.

[0089] Step 200, the optical output end face arranged along the waveguide N r Each detection unit independently performs photoelectric conversion on the second-band fluorescence signal coupled to itself obtained in step 100, and outputs... N r The path receives branch signals, where N r ≥2.

[0090] In step 200, each detection unit corresponds to a different light-emitting position on the waveguide light-emitting end face. Optionally, a microlens disposed between the waveguide light-emitting end face and the detection unit focuses the second-band fluorescence signal output at the corresponding position onto the photosensitive surface of the corresponding detection unit. Each detection unit serves as a receiving sub-aperture unit and retains an independent electrical signal output, so that subsequent processing can utilize the differences in receiving power, channel response, time delay, and distortion state corresponding to different light-emitting positions.

[0091] In step 300, the multi-channel analog front-end amplifies, impedance-matches, and adjusts the gain of each receiving branch signal output in step 200, and keeps each receiving branch signal independent in the analog domain. Subsequently, the multi-channel synchronous acquisition module synchronously samples each receiving branch signal based on the same sampling clock to obtain multiple digitized receiving branch signals.

[0092] In step 300, each detection unit is connected to an analog signal processing channel. The multi-channel analog front-end directly mixes the outputs of different detection units in parallel before channel quality evaluation. The multi-channel synchronous acquisition module can use a multi-channel oscilloscope or an acquisition circuit with multi-channel synchronous analog-to-digital conversion capability. In this embodiment, a multi-channel oscilloscope is used to synchronously acquire each signal, and the sampled data is input into a computer for offline processing.

[0093] Step 400 involves performing channel-level processing on the multiple digitized received branch signals obtained in step 300 to obtain the demodulated symbols of each channel and the channel quality of each received channel. Specifically, step 400 includes steps 410 to 440.

[0094] Step 410: Synchronize and normalize the gain of each digitized receiving branch signal. Synchronization is used to determine the starting position of data blocks, training sequences, or discrete polyphonic symbols in each receiving branch, and gain normalization is used to adjust the signal amplitude of different receiving branches to a range suitable for neural network equalization and subsequent discrete polyphonic demodulation.

[0095] Step 420: Before merging the multiple signals, perform waveform-level equalization based on neural networks on each of the received branch signals that have been synchronized and normalized in gain to obtain the compensated signals of each channel.

[0096] In step 420, the neural network equalizer based on the bidirectional gated recurrent unit takes multiple consecutive received sample values ​​in the neighborhood of the sampling time to be equalized as input, extracts the forward time correlation features and backward time correlation features of the received waveform, and outputs the compensated signal sample value corresponding to the sampling time to be equalized. The waveform-level equalization is used to reduce the power-related gain compression caused by the finite number of micro-units in the silicon photomultiplier tube, as well as the dynamic waveform distortion with time memory characteristics caused by the micro-unit recovery process.

[0097] Step 430: Discrete multi-tone demodulation is performed on each compensated signal obtained in step 420 to obtain each demodulated symbol. The discrete multi-tone demodulation includes serial-to-parallel conversion, removal of cyclic prefix, fast Fourier transform, downsampling, and zero-forcing equalization. When the transmitter adopts a Hermitian symmetric structure, the downsampling includes extracting the non-redundant effective subcarriers corresponding to the Hermitian symmetric structure after the fast Fourier transform.

[0098] Step 440: Using the pilot signal, the channel response amplitude, channel response power, or signal-to-noise ratio of each receiving channel after waveform-level equalization and zero-forcing equalization are estimated to obtain the non-negative channel quality index for each receiving channel. Since the channel quality is estimated after waveform-level equalization is performed on each receiving branch, the impact of gain compression and memory effects in different branches on the channel quality comparison can be reduced.

[0099] Step 500: Determine the combining weight of each receiving channel based on the channel quality obtained in step 440, and perform weighted combining of each demodulated symbol obtained in step 430 in the symbol field to obtain the combined output.

[0100] Specifically, the channel quality indicators of each receiving channel can be normalized according to the merging weight calculation formula described in Example 1 to obtain the corresponding merging weights. Then, the multiple demodulated symbols corresponding to the same discrete multi-tone symbol and the same subcarrier can be weighted and summed according to the symbol domain weighted merging formula described in Example 1. Through this process, the receiving branch with higher channel quality contributes more to the merged symbols, while the receiving branch with lower channel quality contributes less.

[0101] When the location or angle of incidence changes, the channel quality corresponding to different detection units may change. For example, when the incident spot is closer to a local location on the waveguide end face, one or more detection units corresponding to that location may obtain higher received power, while other detection units may obtain relatively lower received power. Step 500 does not pre-fix the contribution ratio of each receiving branch, but instead allocates a combining weight based on the channel quality of each receiving branch after compensation and demodulation, thereby utilizing the differences and complementarities formed between different receiving branches as the incident conditions change.

[0102] Step 600 involves performing parallel-to-serial conversion and demapping on the merged output obtained in step 500 to recover the original data signal from the transmitting end. Optionally, the bit error rate, achievable data transmission rate, or other communication performance indicators can be calculated based on the recovery result.

[0103] Steps 100 to 600 above form a complete physical signal processing chain: First, the fluorescent antenna collects, performs wavelength conversion, and spatial redistributes the visible light communication signal transmitted in free space. Then, multiple detection units perform photoelectric conversion and independent analog processing respectively. Next, the synchronously sampled physical received waveform is subjected to nonlinear and memory effect compensation, discrete multi-tone demodulation, and channel quality estimation path by path. Finally, weighted merging is performed in the symbol domain to recover the transmitting end data. Therefore, the object processed by the method is an electrical signal with a clear physical meaning generated by the optical communication receiving link, rather than abstract data detached from the communication receiving equipment.

[0104] Example 3 This embodiment provides a visible light communication system, including a transmitter and an integrated array visible light communication receiver based on large-area light collection and light field redistribution as described in Embodiment 1. The structure of the receiver is shown in Figure 1, and its three-dimensional structure relationship can be seen in Figure 3.

[0105] The transmitting end is used to modulate the data to be transmitted onto the first-band signal light using discrete multi-tone modulation, and then transmits the modulated first-band signal light into free space. For example, the original bit sequence sequentially undergoes bit loading, constellation mapping, frequency domain pre-equalization, Hermitian symmetry construction, inverse fast Fourier transform, and cyclic prefix addition to generate a time-domain transmission waveform. This time-domain transmission waveform, after digital-to-analog conversion and electrical amplification, drives the visible light transmitter, enabling the first-band signal light to carry the data to be transmitted.

[0106] The first-band signal light, after being transmitted in free space, is incident on the fluorescent antenna of the receiver. The receiver performs the following processes on the first-band signal light according to the method described in Example 2: receiving, photoelectric conversion, per-channel waveform level equalization, per-channel discrete multi-tone demodulation, channel quality estimation, symbol domain weighted combining, and demapping, to recover the data to be transmitted.

[0107] In the test system built in this embodiment, the transmitting end uses a visible light laser to output modulated signal light, and the receiving end uses a silicon photomultiplier tube array with multiple detection units. The output of each detection unit is synchronously sampled by a multi-channel oscilloscope, and the sampled data is processed offline by a computer. The signal processing flow, optical path connection relationship, and board-level integrated receiver assembly arrangement of the transmitting and receiving ends in the test system are shown in Figure 4. The communication performance of the receiver is tested by changing the incident light spot position, spot size, incident angle, free space transmission distance, and ambient light intensity.

[0108] As shown in Figure 5, when the incident light spot is displaced in different directions along the light-receiving area of ​​the fluorescent antenna, the frequency response and normalized time-domain impulse response of the corresponding receiving branch change accordingly. This demonstrates that changes in the position of the incident light spot create differences in bandwidth, response shape, and distortion between different receiving branches. These differences constitute the basis for subsequent path-by-path compensation and channel quality-weighted combining.

[0109] When testing the spatial reception capability of the receiver, the incident light spot is moved within a predetermined performance evaluation area of ​​the fluorescent antenna, and the achievable data transmission rate at different positions is measured. When only one detector unit is used, the reception rate may decrease rapidly after the light spot moves away from the emission position corresponding to that detector unit. When multiple detector units are used and symbol-domain weighted combining is performed, detector units at different positions can receive the second-band fluorescent signal propagating through the planar waveguide, and participate in combining according to their respective channel quality, thereby improving the uniformity of the data transmission rate within the evaluation area. Figure 6 shows the achievable data transmission rate distribution at various measurement points within the evaluation area under different incident light spot sizes for the multi-receiver branch weighted combining mode, the single center detector unit mode, and the single edge detector unit mode. Under incident light spot diameters of 2.5 mm, 5 mm, and 10 mm, the rate-area product corresponding to the multi-receiver branch synthetic aperture receiving mode is 3.45 Gb / s·cm. 2 4.49 Gb / s·cm 2 and 4.134 Gb / s·cm 2 The area of ​​the performance evaluation region is 3.0 cm². 2 Correspondingly, the average achievable data transmission rates within the evaluation area are approximately 1.15 Gb / s, 1.50 Gb / s, and 1.38 Gb / s, respectively. The average rate-area product for the single receiver branch mode is 1.88 Gb / s·cm. 2 2.69 Gb / s·cm 2 and 2.31 Gb / s·cm 2 The corresponding average achievable data transmission rates are approximately 0.63 Gb / s, 0.90 Gb / s, and 0.77 Gb / s, respectively. Compared to the single receiver branch mode, the rate-area product of the multi-receiver branch synthetic aperture receiver mode is improved by approximately 83%, 67%, and 79%, respectively. Test results show that, under the corresponding test conditions, the rate-area product of the multi-receiver branch synthetic aperture receiver mode is higher than that of the single receiver branch mode. The rate-area product is determined by multiplying the average achievable data transmission rate at each measurement point within a predetermined performance evaluation area by the area of ​​that performance evaluation area, rather than by multiplying the highest data transmission rate at a single point by the total area of ​​the fluorescent antenna.

[0110] When testing the receiver's incident angle adaptability, the transmit power, spot size, transmission distance, receiver parameters, and signal processing method were kept essentially constant. The incident beam was gradually deflected relative to the normal direction of the receiving surface, and the achievable data transmission rate at each incident angle was measured or estimated. Using the maximum achievable data transmission rate within the angular scanning range as a benchmark, the positive and negative incident angles corresponding to a data transmission rate reduction to approximately half of the maximum value were defined as the negative 3 dB data rate boundary angles. Test results show that, under the corresponding test conditions and evaluation methods, the receiver can obtain a relatively wide negative 3 dB data rate receiving field of view. Figure 7 shows g and h, respectively, the distribution of achievable data transmission rates with incident angles corresponding to different incident spot sizes in single-receiver branch mode and multi-receiver branch weighted combining mode. When only one detector unit is used, the maximum achievable data transmission rate is approximately 1.5 Gb / s for an incident spot diameter of 2.5 mm, and slightly higher than 1 Gb / s for an incident spot diameter of 10 mm. The achievable data transmission rate only shows a significant decrease when the absolute value of the incident angle exceeds 60°, and it can still maintain a data transmission rate of several hundred Mb / s even at an incident angle of approximately 80°. When all receiver branches are used and weighted combining is performed, the maximum achievable data transmission rate increases to approximately 1.93 Gb / s, and the maximum achievable data transmission rate for an incident spot diameter of 10 mm also reaches approximately 1.8 Gb / s. Under all three incident spot diameter conditions, the achievable data transmission rate remains above 1.5 Gb / s when the absolute value of the incident angle is no greater than 60°. The test results show that, under the corresponding test conditions, the achievable data transmission rate of the multi-receiver branch weighted combining mode is generally higher than that of the single-receiver branch mode at various incident angles, and it can maintain a high achievable data transmission rate over a wide range of incident angles. Furthermore, under the three incident spot diameter conditions, according to the evaluation method used in this test, the half field of view of the receiver with a data rate of -3 dB is estimated to be between 75° and 80°, that is, the receiving field of view with a data rate of -3 dB is approximately ±75° to ±80°.

[0111] Furthermore, to evaluate the synergistic effect between path-by-path waveform equalization and channel quality-weighted combining, the following post-processing configurations were compared for synchronously sampled data obtained under the same test conditions: equal-gain combining without neural network waveform equalization; equal-gain combining with neural network waveform equalization; channel quality-based weighted combining without neural network waveform equalization; and path-by-path neural network waveform equalization followed by channel quality-based weighted combining after compensation. This comparative test was conducted under conditions of a free-space transmission distance of 1.75 m and an incident light spot diameter of 2.5 mm. Under the corresponding test conditions, the achievable data transmission rate (ATR) for the configuration of performing per-path neural network waveform equalization followed by channel quality-based weighted combining after compensation is 1.7578 Gb / s; the achievable ATR for the configuration of using channel quality-based weighted combining without neural network waveform equalization is 1.6641 Gb / s; the achievable ATR for the configuration of equal-gain combining with neural network waveform equalization is 1.6016 Gb / s; and the achievable ATR for the configuration of equal-gain combining without neural network waveform equalization is 1.5488 Gb / s. The test results show that both waveform equalization and quality-based weighted combining can improve the achievable ATR based on the baseline configuration. The configuration combining per-path waveform equalization and compensated quality-based weighted combining achieves the highest achievable ATR among the four configurations, indicating that the two have a synergistic improvement effect under the corresponding test conditions.

[0112] In another set of ambient light tests, four configurations were set up: single receiver branch without waveform level equalization, single receiver branch with waveform level equalization, multi-receiver branch weighted combination without waveform level equalization, and multi-receiver branch weighted combination with waveform level equalization. The effects of the number of receiver sub-apertures and waveform level equalization were differentiated. Test results show that the configuration with multi-receiver branch weighted combination and waveform level equalization achieved the highest achievable data transmission rate among the four configurations, and waveform level equalization showed additional improvement under the multi-receiver branch quality-weighted configuration.

[0113] The achievable data transmission rates of the four configurations under different ambient light intensities are shown in Figure 7d. Test results indicate that, under the corresponding test conditions, the achievable data transmission rate of each configuration decreases with increasing ambient light intensities. The configuration with multiple receiver branches weighted summation and waveform-level equalization maintains the highest achievable data transmission rate among the four configurations across the entire test illumination range. Figure 7b further illustrates the modulation order distribution and corresponding signal-to-noise ratio distribution of each effective subcarrier after channel detection and bit power loading under different ambient light intensities. It can be seen that as ambient light intensities increase, the overall signal-to-noise ratio of each effective subcarrier decreases. The bit power loading process correspondingly reduces the modulation order of some subcarriers, enabling the system to maintain link error performance even when ambient light conditions change.

[0114] Figure 7f compares the transmission rate reduction of the receiving method with the fluorescent antenna and the receiving method without the fluorescent antenna, where the detection unit receives directly, under different ambient light intensities. The test results show that, under the corresponding test conditions, as the ambient light intensities increase, the transmission rate reduction without the fluorescent antenna is significantly greater than that with the fluorescent antenna. This indicates that the wavelength selective switching of the fluorescent antenna, combined with the filtering effect of the filter layer, helps to suppress the adverse effects of ambient background light on the receiving performance.

[0115] When testing the system's performance at different transmission distances, the free-space transmission distance between the transmitter and receiver was varied within a predetermined range, and the achievable data transmission rate and bit error rate were measured for different incident spot sizes. As shown in Figure 7a, under the corresponding test conditions, the achievable data transmission rate for each spot size generally decreased with increasing transmission distance. Specifically, when the incident spot diameter was 5 mm, the achievable data transmission rate decreased from approximately 1.93 Gb / s at 0.75 m to approximately 1.86 Gb / s at 2.0 m, a decrease of approximately 3.8%. Throughout the entire test distance range, the total decrease in achievable data transmission rate for incident spot diameters of 2.5 mm, 5 mm, and 10 mm was approximately 220 Mb / s, 75.2 Mb / s, and 191 Mb / s, respectively, with the achievable data transmission rate for each spot size remaining at a relatively high level. As shown in Figure 7c, under the condition of using the waveform-level equalization, the bit error rate corresponding to each spot size is lower than the predetermined forward error correction bit rate threshold throughout the entire test distance range; while without the waveform-level equalization, the corresponding bit error rate is higher than the threshold. Figure 7e shows the achievable data transmission rate and bit error rate obtained from multiple repeated tests under the same test conditions. The test results are basically consistent, indicating that the receiver has good repeatability and stability. The test results under different transmission distances and ambient light conditions also show that the fluorescent antenna can receive signal light with varying incident positions, incident angles, or transmission distances within a certain range. Multiple independent receiving branches can retain the channel differences corresponding to different light output positions. Path-by-path waveform-level equalization and weighted combining of the compensated symbol domain can utilize these differences to improve the reception results. The test results are used to illustrate the feasibility and performance of this embodiment under the corresponding experimental conditions and should not be construed as an absolute limitation on the performance of the receiver under other device parameters, other light sources, or other transmission environments.

[0116] To further illustrate the feasibility of the foregoing embodiments, a set of exemplary device parameters and signal parameters used in the aforementioned test system are given below. It should be understood that the following parameters are merely examples, and the implementation of this application is not limited to the specific parameters described below.

[0117] In this example, the integrated silicon photomultiplier tube array detector 20 includes four detection units 21, namely... Each detector unit 21 exemplarily employs a MicroFJ-30035-TSV commercial silicon photomultiplier tube device. The photosensitive surface size of a single detector unit is 3.07 mm × 3.07 mm, with 5676 micro-units, each 35 μm in size, a fill factor of 75%, and a recovery time constant of approximately 45 ns. The operating bias voltage in the experiment is 27.0 V. The four detector units 21 are arranged laterally in a 1×4 parallel configuration, with a center-to-center spacing of 5 mm between adjacent detector units. They are aligned with the waveguide end face 102 at the lower end of the fluorescent antenna 10, ensuring that the total receiving width of the four detector units 21 in the width direction of the waveguide end face matches the width of the waveguide end face 102.

[0118] In this example, the fluorescent antenna 10 has a planar dimension of 20 mm × 20 mm and a thickness of 5 mm. After taking into account the filtering effect of the filter layer 14, the effective receiving band of the fluorescent antenna 10 for incident light is 380 nm to 525 nm, and the emission peak wavelength of the second band fluorescent signal is about 515 nm, which matches the response band of the detection unit 21.

[0119] In this example, the discrete multitone signal at the transmitter is generated according to the following parameters: the sampling rate of the arbitrary waveform generator is set to 2 GSa / s, the number of inverse fast Fourier transform points (i.e., the total number of subcarriers) is 1040, of which there are 512 effective subcarriers and 8 empty subcarriers. The cyclic prefix length is 20% of the length of the discrete multitone symbol. The modulation order of each effective subcarrier is determined according to the bit power loading result after channel probing. The upsampling factor at the transmitter is 2.

[0120] Accordingly, in the receiver processing of this example, the time-domain waveform obtained by synchronous sampling is first synchronized, gain normalized, and subjected to the aforementioned waveform-level equalization. After anti-aliasing filtering, it is decimated at twice the decimation rate to restore the sampling rate of the received waveform to the discrete multi-tone baseband sampling rate. Subsequently, the decimated waveform is subjected to discrete multi-tone symbol segmentation and cyclic prefix removal, and a 1040-point fast Fourier transform is performed. According to the Hermitian conjugate symmetric mapping relationship used by the transmitter, the single-sided spectrum carrying independent information is extracted from the fast Fourier transform output, the conjugate symmetric mirror spectrum is discarded, and 8 empty subcarriers are removed, thereby recovering the received symbols on 512 effective subcarriers. This completes the downsampling in the discrete multi-tone demodulation described above, and zero-forcing equalization is performed on the received symbols.

[0121] In this example, as a specific implementation of the sliding window input method described above, the neural network equalizer based on a bidirectional gated recurrent unit employs a two-level sliding window: 77 continuously received sample values ​​are input at each time step, for a total of 15 time steps, with sliding between adjacent time steps. Therefore, the original waveform length covered by a single forward computation is 77 + 15 - 1 = 91 sample points. The bidirectional gated recurrent unit layer is single-layered, with a hidden state dimension of 77 in each direction. The training batch size is 128, and the Adam optimization algorithm is used with a learning rate on the order of magnitude [missing information]. The first 30% of the collected data is used as training data, and the remaining 70% is used as validation data; the four receiving branches share the same set of network parameters.

[0122] In the spatial reception capability test of this example, the size of the performance evaluation area is 15 mm × 20 mm, the spatial step size of the measurement points is 5 mm, and there are 12 actual measurement points. After spatial interpolation of the actual measurement results, a total of 63 evaluation points (7 × 9) are obtained. The rate-area product is determined by multiplying the average data transmission rate achievable by the 63 evaluation points by the area of ​​the performance evaluation area.

[0123] In the incident angle adaptability test and processing configuration comparison test of this example, the emitted optical power was 20 mW, the free space transmission distance was 0.75 m, the incident spot diameter was adjusted to 2.5 mm, 5 mm, and 10 mm, and the angle scanning step size was 15°. The comparison test of the four processing configurations used an incident spot with a diameter of 2.5 mm and was based on the same set of synchronously sampled data, implemented through different offline post-processing methods. The achievable data transmission rate in each test was uniformly estimated according to the following formula: in, To achieve the achievable data transmission rate; The average modulation order of the effective subcarriers (in terms of the number of bits carried per symbol per subcarrier) is determined based on the bit power loading results; The sampling rate set for the arbitrary waveform generator is 2 GSa / s; This is the number of effective subcarriers, i.e., 512; This represents the total number of subcarriers, which is 1040. This is the upsampling factor at the transmitting end, which is 2.

[0124] The above exemplary parameters are used to illustrate a specific implementation of the foregoing embodiments under corresponding experimental conditions. Those skilled in the art can make corresponding adjustments to the above values ​​based on the actual devices, optical paths, and signal parameters used.

[0125] The above embodiments have the following technical effects: By utilizing the larger light-receiving area of ​​the fluorescent antenna, fluorescence conversion, and planar waveguide propagation, the first-band signal light incident at different positions and angles is reorganized into second-band fluorescent signals distributed and output from different positions on the waveguide end face, which helps to expand the usable range of incident positions and incident angles; by setting multiple independent detection units along the waveguide end face and connecting each detection unit to an independent analog signal processing channel and a synchronous acquisition channel, multiple receiving sub-apertures are formed without increasing the photosensitive area of ​​a single high-speed detection unit, while preserving the differences in received power, channel response, and distortion state corresponding to different light-emitting positions; and by performing waveform-level equalization based on neural networks before combining multiple signals. This approach can reduce the power-related gain compression and memory effect caused by the finite number of micro-units in silicon photomultiplier tubes and the recovery process, making each receiving branch more suitable for discrete multi-tone demodulation and channel quality estimation. By performing waveform-level equalization, discrete multi-tone demodulation, and channel quality estimation on each path, and then weighting and combining them in the symbol domain according to channel quality, different receiving branches can participate in the combining according to their compensated effective receiving state, which helps to improve the uniformity of data transmission rate and receiving stability under conditions of varying spot position or incident angle. Experimental results also show that, under the corresponding test conditions, the achievable data transmission rate obtained by jointly using path-by-path waveform-level equalization and symbol domain weighted combining based on compensated channel quality is higher than the results obtained by using only one of the processing measures.

[0126] It should be noted that in this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this application, if a reference is made to performing an action based on an element, it means performing the action at least based on that element, including two cases: performing the action only based on that element, and performing the action based on that element and other elements. Expressions such as "multiple," "repeatedly," and "various" include two, two times, two kinds, and more than two, more than two times, and more than two kinds. The above descriptions are only partial embodiments of this application and are not intended to limit this application. Those skilled in the art can make corresponding changes or substitutions to the above embodiments without departing from the technical concept of this application.

Claims

1. An integrated array visible light communication receiver based on large-area light collection and light field redistribution, characterized in that, include: A fluorescent antenna is disposed at the incident end of the receiver and has a light-receiving area and a waveguide light-out end face. The fluorescent antenna is used to receive first-band signal light incident at different positions and angles in the light-receiving area, convert the first-band signal light into a second-band fluorescent signal, and make the second-band fluorescent signal propagate in multiple directions in the planar waveguide structure of the fluorescent antenna and then output it from different positions of the waveguide light-out end face. An integrated silicon photomultiplier tube array detector, comprising arrays arranged along the waveguide end face. N r One detection unit, N r ≥ 2; Each of the detection units is optically coupled to different light-emitting positions of the waveguide light-emitting end face, and each independently converts the second band fluorescence signal coupled to itself into a receiving branch signal; Multi-channel analog front end, including with N r Each of the detection units is connected to N r Each analog signal processing channel amplifies the corresponding received branch signal and keeps each received branch signal independent of the others in the analog domain. A multi-channel synchronous acquisition module is used to synchronously sample the receiving branch signals output by the multi-channel analog front end based on the same sampling clock, so as to obtain multiple digitized receiving branch signals. The digital signal processing module is used to: synchronize and normalize the gain of each of the digitized receiving branch signals; before merging the multiple signals, perform waveform-level equalization based on neural networks on each of the synchronized and normalized receiving branch signals to compensate for the nonlinear distortion and memory effect of each receiving channel, and obtain the compensated signals; and perform discrete multi-tone demodulation on each of the compensated signals to obtain the demodulated symbols, wherein the discrete multi-tone demodulation includes serial-to-parallel conversion, removal of cyclic prefix, fast Fourier transform, downsampling, and zero-forcing equalization. The channel quality of each receiving channel is estimated using pilot signals; The combining weight of each receiving channel is determined based on the channel quality. The demodulated symbols of each channel are weighted and combined in the symbol domain to obtain the combined output. The combined output is then subjected to parallel-to-serial conversion and demapping to recover the original data signal from the transmitting end.

2. The receiver according to claim 1, characterized in that, The fluorescent antenna includes an incident window, a fluorescence conversion layer, the planar waveguide structure, a filter layer, a reflective layer, and an encapsulation structure; The first band signal light enters the fluorescent antenna through the incident window, is absorbed by the fluorescent conversion layer and converted into a second band fluorescent signal, and the band of the second band fluorescent signal matches the response band of the detection unit; The second-band fluorescence signal propagates in multiple directions within the planar waveguide structure and couples from different positions on the waveguide end face to... N r The aforementioned detection unit.

3. The receiver according to claim 1 or 2, characterized in that, The photosensitive area of ​​a single detection unit is smaller than the light-receiving area of ​​the light-receiving region; N r The detection units are arranged in parallel along the waveguide end face, and the number and arrangement of the detection units are determined according to the size of the waveguide end face. N r The total receiving width of each of the detection units is matched with the width of the waveguide end face, so that the second band fluorescence signals distributed and output from different positions of the waveguide end face are coupled to different detection units respectively.

4. The receiver according to claim 1 or 2, characterized in that, It also includes a microlens array, which is disposed between the waveguide end face and the integrated silicon photomultiplier tube array detector, wherein each microlens in the microlens array and N r Each of the detection units corresponds to a specific microlens, and each microlens converges the second-band fluorescence signal output from the corresponding position of the wavelength-derived light-emitting end face to the photosensitive surface of the corresponding detection unit.

5. The receiver according to claim 4, characterized in that, Each of the analog signal processing channels also performs impedance matching and gain adjustment on the corresponding receiving branch signal; The integrated silicon photomultiplier tube array detector and the multi-channel analog front end are disposed on the same printed circuit board, and the microlens array is mounted on an optical bracket that is fixed relative to the printed circuit board. The microlens array, the integrated silicon photomultiplier tube array detector, and the multi-channel analog front end constitute a board-level integrated receiving component.

6. The receiver according to claim 1 or 2, characterized in that, The waveform-level equalization is performed by a neural network equalizer based on a bidirectional gated cyclic unit. The neural network equalizer takes multiple consecutive received sample values ​​in a sliding window in the neighborhood of the sampling time to be equalized as input, extracts the forward time correlation features and backward time correlation features of the received waveform through the bidirectional gated cyclic unit, and outputs the compensated signal sample value corresponding to the sampling time to be equalized.

7. The receiver according to claim 6, characterized in that, Each of the aforementioned detection units is composed of multiple micro-units, and the micro-units recover their gain after an avalanche response through a recovery process with a recovery time constant; The nonlinear distortion and memory effect include at least: gain compression caused by the limited number of micro-units, in which the output amplitude deviates from the linear proportional relationship as the incident light power increases; and dynamic waveform distortion with time memory characteristics caused by the recovery process of the micro-units, in which the current output amplitude is affected by the previous input waveform.

8. The receiver according to claim 1 or 2, characterized in that, Let the first i The demodulated symbol obtained after the received branch signal is processed by the waveform level equalization and the discrete multi-tone demodulation is: Ŝ i ( k , m The first, estimated using the pilot signal i The non-negative channel quality index for each receiving channel is: q i ,and If the combined weight is greater than 0, then the combined weight is... w i Determined according to the following formula: The digital signal processing module performs a weighted merging of the demodulated symbols in the symbol domain according to the following formula: in, i and j These are all sequence numbers of the receiving channel. i , j = 1, 2, ⋯, N r ; N r The number of the detection units; k Subcarrier sequence number; m For discrete polyphonic symbols; Ŝ i ( k , m ) is the first i The first branch signal obtained after the waveform level equalization and discrete multi-tone demodulation of the received branch signal is... m In the discrete polyphonic symbols, the first k Demodulated symbols on each subcarrier; q i In order to utilize the pilot signal according to the first i The non-negative channel quality index is determined by the channel response amplitude or signal-to-noise ratio of each receiving channel after the waveform level equalization and the zero-forcing equalization. w i For the first i The merging weight of each receiving channel; Ŝ ( k , m ) is the first in the merged output m In the discrete polyphonic symbols, the first k Combined symbols on each subcarrier.

9. A visible light communication receiving method based on an integrated array using large-area light collection and light field redistribution, characterized in that, The method is implemented based on the receiver of any one of claims 1 to 8, and includes: Step 100: The fluorescent antenna receives first-band signal light incident at different positions and angles in the light-receiving area, converts the first-band signal light into a second-band fluorescent signal, and causes the second-band fluorescent signal to propagate in multiple directions in the planar waveguide structure before being distributed and output from different positions on the waveguide end face. Step 200: Each of the aforementioned detection units independently performs photoelectric conversion on the second band fluorescence signal coupled to itself obtained in step 100, and outputs... N r The path receives branch signals; Step 300: The multi-channel analog front end amplifies each of the received branch signals output in step 200, and the multi-channel synchronous acquisition module synchronously samples each of the received branch signals based on the same sampling clock to obtain multiple digitized received branch signals; Step 400: Synchronize and normalize the digitized receiving branch signals obtained in Step 300 respectively; before combining the multiple signals, perform waveform-level equalization based on neural networks on each of the synchronized and normalized receiving branch signals to compensate for the nonlinear distortion and memory effect of each receiving channel, and obtain the compensated signals; perform discrete multi-tone demodulation on each of the compensated signals to obtain the demodulated symbols, and use the pilot signal to estimate the channel quality of each receiving channel. The discrete multi-tone demodulation includes serial-to-parallel conversion, removal of cyclic prefix, fast Fourier transform, downsampling and zero-forcing equalization. Step 500: Determine the combining weight of each receiving channel based on the channel quality obtained in step 400, and perform weighted combining of each demodulated symbol obtained in step 400 in the symbol domain to obtain the combined output; Step 600: Perform parallel-to-serial conversion and demapping on the merged output obtained in step 500 to recover the original data signal from the transmitting end.

10. A visible light communication system, characterized in that, Includes a transmitter and a receiver as described in any one of claims 1 to 8; The transmitting end is used to modulate the data to be transmitted onto the first band signal light in a discrete multi-tone modulation method, and to transmit the modulated first band signal light into free space. The receiver is used to receive the first band signal light that has been transmitted in free space and incident on the light-receiving area, and to recover the data to be transmitted.

Images