An ultra-hundred-gigabit optical network neural bus integrating sensing and computing and a running method thereof

Abstract

The present application relates to the field of optical communication technology, and more particularly to a super-hundred G optical network nerve bus integrating sensing, computing and running method, the super-hundred G optical network nerve bus integrating sensing, computing and running method comprises a flexible optical waveguide circuit board, a plurality of bionic neuron optical engine modules, a power management module and an intelligent optical routing scheduling module; the plurality of bionic neuron optical engine modules are connected with the flexible optical waveguide circuit board respectively, the power management module is connected with the flexible optical waveguide circuit board, and the intelligent optical routing scheduling module is connected with the flexible optical waveguide circuit board; the present application is a super-hundred G optical network nerve bus integrating sensing, computing and running method supporting dead angle free environment sensing, zero sensing delay large bandwidth transmission, nerve reflex level real-time response and cell embedding type conformal deployment function, and the bus has the characteristics of high integration, good universality, high intelligence, good reliability, low power consumption, low time delay, low cost, easy expansion, easy iteration, easy deployment and the like.

Description

Technical Field

[0001] This invention relates to the field of optical communication technology, and in particular to a neural bus for ultra-100G optical networks integrating communication, sensing, and computing, and its operation method. Background Technology

[0002] The 100G+ optical network bus features high bandwidth, high integration, low latency, anti-interference, and lightweight design. It is widely used in unmanned platforms such as drones, unmanned vehicles, unmanned boats, and space-based satellites, and is a key development direction for the next generation of optical communication technology.

[0003] However, existing 100G+ optical network buses only support data transmission. When applied to unmanned platforms, they require not only the addition of numerous data sensing terminals and edge computing terminals, but also a large number of signal interconnection devices. This stacking of devices not only consumes a large amount of valuable payload space on the unmanned platform, but also poses a huge challenge to achieving lightweight, low-power, highly integrated, highly reliable, and universal unmanned platforms. This has become a key pain point and bottleneck problem restricting the promotion and application of 100G+ optical network neural buses on unmanned platforms.

[0004] For the reasons mentioned above, there is an urgent need for a super-100G optical network neural bus that integrates sensing, computing, and communication to meet the application requirements of various unmanned platforms such as unmanned vehicles, unmanned boats, drones, and space-based satellites for multi-dimensional information perception, real-time near-sensing computing, and low-latency broadband transmission. Summary of the Invention

[0005] The purpose of this invention is to provide a super-100G optical network neural bus and its operation method that integrates sensing, computing, and communication, which can meet the application requirements of unmanned platforms for multi-dimensional information perception, real-time near-sensing computing, and low-latency broadband transmission.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a neural bus for ultra-100G optical networks that integrates sensing, computing, and communication, including a flexible optical waveguide circuit board, multiple biomimetic neuron optical engine modules, a power management module, and an intelligent optical routing scheduling module.

[0007] Multiple biomimetic neuron light engine modules are connected to the flexible light waveguide circuit board, the power management module is connected to the flexible light waveguide circuit board, and the intelligent light routing scheduling module is connected to the flexible light waveguide circuit board.

[0008] The system comprises six bionic neuron light engine modules. The flexible optical waveguide circuit board has a flexible electrical connection point array and a MT connector. The electrical signals of the six bionic neuron light engine modules are interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array using a press-fit method. The optical signals of the six bionic neuron light engine modules are interconnected with the flexible optical waveguide circuit board via the MT connector using a plug-in method. The power management module is interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array using a press-fit method. The intelligent optical routing scheduling module is interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array using a press-fit method.

[0009] The bionic neuron optical engine module includes a signal processing unit, a high-speed photoelectric conversion unit, a multi-rate photoelectric conversion unit, a clock unit, an MT optical port plug-in unit, a flexible pressure contact unit, a cross-mode sensor array unit, an edge computing CPU unit, and a high-speed data cache unit.

[0010] The high-speed photoelectric conversion unit, the multi-rate photoelectric conversion unit, and the clock unit are respectively connected to the signal processing unit; the MT optical port connector unit is respectively connected to the high-speed photoelectric conversion unit and the multi-rate photoelectric conversion unit via optical fibers; the flexible crimp contact unit, the cross-mode sensor array unit, the edge computing CPU unit, and the high-speed data cache unit are respectively connected to the signal processing unit.

[0011] The signal processing unit is composed of an FPGA chip of model FM9VU13PB2104, which is used to perform protocol conversion and interface matching processing on various acquired signals and data.

[0012] The high-speed photoelectric conversion unit is used to complete the electro-optical and photoelectric signal conversion of one 100G or one 200G electrical signal; the multi-rate photoelectric conversion unit is used to complete the electro-optical and photoelectric signal conversion of one 155M or one 622M or one 1.25G or one 2.5G or one 10G electrical signal.

[0013] The clock unit is used to provide the signal processing unit with four high-precision, low-jitter clocks at frequencies of 156.25M, 155.52M, 174M, and 19.44M; the MT optical port plug-in unit is used to provide an optical coupling interface for optical signal transmission between the bionic neuron optical engine module and the flexible optical waveguide circuit board.

[0014] The flexible pressure contact unit is used to provide an electrical connection channel for the electrical signal transmission between the bionic neuron light engine module and the flexible light waveguide circuit board and the unmanned platform; the cross-mode sensor array unit is used to provide the bionic neuron light engine module with real-time sensing information including temperature, tilt angle, angular velocity, acceleration, pressure, deformation, illumination and air pressure.

[0015] The edge computing CPU unit is used to perform real-time edge deployment computing on the data collected and received by the bionic neuron light engine module; the high-speed data cache unit is used to provide the signal processing unit with a high-speed storage space with a capacity of 128G, a throughput rate of 200GB / S, and a latency of 50ns.

[0016] The flexible optical waveguide circuit board provides a physical channel for the optical network neural bus to be conformally deployed on an unmanned platform for hybrid optical and electrical transmission. The power management module converts the 24V secondary power input from the unmanned platform into multiple power sources of 12V, 5V, and 3.3V, which are then output through the flexible optical waveguide circuit board to the six bionic neuron optical engine modules and the intelligent optical routing scheduling module to provide power to the optical network neural bus. The intelligent optical routing scheduling module uses an adaptive intelligent routing algorithm to switch the 12 optical signals from the six bionic neuron optical engine modules to the corresponding bionic neuron optical engine modules.

[0017] Secondly, the present invention also provides a method for operating a neural bus of a super-100G optical network integrating sensing, computing, and communication, comprising:

[0018] S1 Power-on Initialization: After power-on, the functional modules of the integrated inductive computing ultra-100G optical network neural bus are initialized;

[0019] S2 System Self-Check: Automatically detects whether the bionic neuron light engine module, flexible light waveguide circuit board, power management module, and intelligent light route scheduling module are working properly;

[0020] S3 Multidimensional Sensing Data Extraction: Reads the raw data collected by the cross-mode sensor array unit and forms raw data frames in time order. The raw data frames include image data frames, point cloud data frames, waveform data frames, and time-series numerical frames.

[0021] S4 Real-time Edge Computing: Uses CPUs deployed on edge nodes to perform real-time computation on multi-dimensional sensing data extracted by the cross-mode sensor array in the bionic neuron optical engine module;

[0022] S5 Traffic Aggregation: The bionic neuron optical engine module aggregates the large-capacity real-time data from the unmanned platform and the data generated by real-time edge computing into one data stream, which is then used to transmit the optical signal bandwidth of the flexible optical waveguide circuit board to the intelligent optical routing module.

[0023] S6 Homomorphic Encryption: The bionic neuron light engine module completely hides the plaintext and incorporates controllable noise into the ciphertext through public key linear transformation and linear noise superposition, thus completing data encryption.

[0024] S7 error correction coding: The bionic neuron light engine module adds redundant check bits to the original information code elements according to the RS coding rules;

[0025] S8 Intelligent Routing and Switching: Based on the MAP table provided by each bionic neuron optical engine module, it analyzes and generates the optimal data exchange routing matrix table for the integrated sensory computing ultra-100G optical network neural bus in real time. The intelligent optical routing scheduling module automatically completes the routing scheduling of all optical signals based on the data exchange routing matrix table.

[0026] S9 Error Correction Decoding: The bionic neuron optical engine module performs error correction decoding on the received data according to the RS encoding rules;

[0027] S10 Edge Trust Computing: The bionic neuron optical engine module performs edge trust computing on the received broadband high-capacity data in real time to ensure that the data, code, identity, and communication are trustworthy. Data that fails the edge trust computing is isolated and the edge trust computing results are reported to other bionic neuron optical engine modules.

[0028] S11 Traffic Balancing: For broadband, high-capacity data transmitted via edge trust computing, the signal processing unit and high-speed data caching unit of the bionic neuron optical engine module work together to perform traffic balancing, ensuring that data is continuously transmitted to the unmanned platform at a stable rate and avoiding data loss caused by the unmanned platform's inability to handle short-term ultra-large-capacity data.

[0029] This invention discloses a sensing-computing integrated ultra-100G optical network neural bus and its operation method. This ultra-100G optical network neural bus supports seamless environmental perception, zero-latency high-bandwidth transmission, "neural reflex" level real-time response, and "cellular embedding" conformal deployment. It can meet the application requirements of various unmanned platforms such as unmanned vehicles, unmanned surface vessels, drones, and space-based satellites for multi-dimensional information perception, near-sensor real-time computing, and low-latency broadband transmission. This bus features high integration, good versatility, high intelligence, high reliability, low power consumption, low latency, low cost, easy expansion, easy iteration, and easy deployment. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0031] Figure 1This is a schematic diagram of the structure of a super-100G optical network neural bus integrating induction and computing according to the present invention.

[0032] Figure 2 This is a schematic diagram of the structure of the bionic neuron light engine module of the present invention.

[0033] Figure 3 This is a flowchart of steps S1-S6 of a method for operating a super-100G optical network neural bus integrating induction and computing according to the present invention.

[0034] Figure 4 This is a flowchart of steps S7-S11 of the operating method of a super-100G optical network neural bus integrating communication, induction and computing according to the present invention.

[0035] 1-Flexible optical waveguide circuit board, 2-Bionic neuron optical engine module, 3-Power management module, 4-Intelligent optical route scheduling module, 21-Signal processing unit, 22-High-speed photoelectric conversion unit, 23-Multi-rate photoelectric conversion unit, 24-Clock unit, 25-MT optical port plug-in unit, 26-Flexible crimp contact unit, 27-Cross-mode sensor array unit, 28-Edge computing CPU unit, 29-High-speed data cache unit. Detailed Implementation

[0036] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0037] Firstly, please refer to Figures 1-2 This invention provides a super-100G optical network neural bus integrating sensing, computing, and communication, comprising a flexible optical waveguide circuit board 1, multiple biomimetic neuron optical engine modules 2, a power management module 3, and an intelligent optical routing scheduling module 4; the biomimetic neuron optical engine module 2 includes a signal processing unit 21, a high-speed photoelectric conversion unit 22, a multi-rate photoelectric conversion unit 23, a clock unit 24, an MT optical port plug-in unit 25, a flexible crimp contact unit 26, a cross-mode sensor array unit 27, an edge computing CPU unit 28, and a high-speed data cache unit 29; the aforementioned solution can meet the application requirements of multi-dimensional information perception, near-sensing real-time computing, and low-latency broadband transmission for unmanned platforms.

[0038] In this specific embodiment, multiple biomimetic neuron optical engine modules 2 are connected to the flexible optical waveguide circuit board 1, the power management module 3 is connected to the flexible optical waveguide circuit board 1, and the intelligent optical routing scheduling module 4 is connected to the flexible optical waveguide circuit board 1. The biomimetic neuron optical engine modules 2, the intelligent optical routing scheduling module 4, the power management module 3, and the flexible optical waveguide circuit board 1 form a "sensory-computing integrated" ultra-100G optical network neural bus. Under the control of the intelligent optical routing scheduling module 4, this "sensory-computing integrated" ultra-100G optical network neural bus achieves cross-modal, blind-spot-free environmental perception, zero-latency, high-bandwidth transmission, "neural reflex" level real-time response, and "cell-embedded" conformal deployment based on biomimetic neuron AI optical engine technology.

[0039] The system comprises six bionic neuron light engine modules 2. The flexible optical waveguide circuit board 1 has a flexible electrical connection point array and an MT connector. The electrical signals of the six bionic neuron light engine modules 2 are interconnected with the flexible optical waveguide circuit board 1 via the flexible electrical connection point array in a press-fit manner. The optical signals of the six bionic neuron light engine modules 2 are interconnected with the flexible optical waveguide circuit board 1 via the MT connector in a plug-in manner. The power management module 3 is interconnected with the flexible optical waveguide circuit board 1 via the flexible electrical connection point array in a press-fit manner. The intelligent optical routing scheduling module 4 is interconnected with the flexible optical waveguide circuit board 1 via the flexible electrical connection point array in a press-fit manner.

[0040] Secondly, the high-speed photoelectric conversion unit 22, the multi-rate photoelectric conversion unit 23, and the clock unit 24 are respectively connected to the signal processing unit 21; the MT optical port connector unit is respectively connected to the high-speed photoelectric conversion unit 22 and the multi-rate photoelectric conversion unit 23 via optical fibers; the flexible pressure contact unit 26, the cross-mode sensor array unit 27, the edge computing CPU unit 28, and the high-speed data cache unit 29 are respectively connected to the signal processing unit 21. In this embodiment of the invention, the six bionic neuron light engine modules 2 are identical small-volume, highly integrated, low-power standardized modules, with a single module volume of 50cm × 40cm × 20cm. Each biomimetic neuron optical engine module 2 includes a signal processing unit 21, a high-speed photoelectric conversion unit 22, a multi-rate photoelectric conversion unit 23, a clock unit 24, an MT optical port connector unit 25, a flexible crimp contact unit 26, a cross-mode sensor array unit 27, an edge computing CPU unit 28, and a high-speed data cache unit 29. The MT optical port connector unit is connected to the high-speed photoelectric conversion unit 22 and the multi-rate photoelectric conversion unit 23 via G653 microbending-resistant bare optical fiber. The signal processing unit 21, the high-speed photoelectric conversion unit 22, the multi-rate photoelectric conversion unit 23, the clock unit 24, the flexible crimp contact unit 26, the cross-mode sensor array unit 27, the edge computing CPU unit 28, and the high-speed data cache unit 29 are connected via high-speed PCB traces. The signal processing unit 21, high-speed photoelectric conversion unit 22, multi-rate photoelectric conversion unit 23, clock unit 24, MT optical port plug-in unit 25, flexible crimping unit, cross-mode sensor array unit 27, edge computing CPU unit 28, and high-speed data cache unit 29 constitute the bionic neuron light engine module 2. Under the unified control of the edge computing CPU unit 28, the bionic neuron light engine module 2 realizes cross-modal, dead-angle-free environmental perception, zero-latency, high-bandwidth transmission, and "neural reflex" level real-time response.

[0041] Meanwhile, the signal processing unit 21 is composed of an FPGA chip of model FM9VU13PB2104, used for protocol conversion and interface matching processing of various acquired signals and data. In this embodiment of the invention, the signal processing unit 21 is composed of the domestically produced high-performance FPGA chip FM9VU13PB2104 from Shanghai Fudan Microelectronics, and is connected to the high-speed photoelectric conversion unit 22, the multi-rate photoelectric conversion unit 23, the cross-mode sensor array unit 27, and the edge computing CPU unit 28 through the high-speed SERDES port of the FPGA. The function of the signal processing unit 21 is to perform protocol conversion and interface matching processing on various acquired signals / data.

[0042] In addition, the high-speed photoelectric conversion unit 22 is used to complete the electro-optical and photoelectric signal conversion of one 100G or one 200G electrical signal; the multi-rate photoelectric conversion unit 23 is used to complete the electro-optical and photoelectric signal conversion of one 155M, one 622M, one 1.25G, one 2.5G, or one 10G electrical signal. In this embodiment of the invention, the high-speed photoelectric conversion unit 22 is a miniaturized 100G / 200G rate adaptive optical module FXSOMK-100 / 200G-I with integrated transceiver, which is connected to the domestic high-performance FPGA chip FM9VU13PB2104 of the signal processing unit 21 through 10 pairs of 25G rate SERDES differential lines. The function of the high-speed photoelectric conversion unit 22 is to complete the electro-optical / photoelectric signal conversion of one 100G or one 200G electrical signal. The multi-rate optoelectronic conversion unit 23 is a miniaturized 155M, 622M, 1.25G, 2.5G, and 10G rate adaptive optical module FXSOMK-155M / 10G-I, which integrates transceiver capabilities. It is connected to the domestically produced high-performance FPGA chip FM9VU13PB2104 of the signal processing unit 21 via a pair of 10G rate SERDES differential lines. The function of the multi-rate optoelectronic conversion unit 23 is to perform electro-optical / optoelectronic signal conversion for one 155M, one 622M, one 1.25G, one 2.5G, or one 10G electrical signal.

[0043] Furthermore, the clock unit 24 provides the signal processing unit 21 with four high-precision, low-jitter clock frequencies: 156.25MHz, 155.52MHz, 174MHz, and 19.44MHz. The MT optical port connector 25 provides an optical coupling interface for optical signal transmission between the bionic neuron optical engine module 2 and the flexible optical waveguide circuit board 1. In this embodiment, the clock unit 24 is composed of a high-precision clock chip XCOF155HB2, which is connected to the domestically produced high-performance FPGA chip FM9VU13PB2104 of the signal processing unit 21 via four pairs of high-speed clock differential lines. The function of the clock unit 24 is to provide the signal processing unit 21 with four high-precision, low-jitter clock frequencies: 156.25MHz, 155.52MHz, 174MHz, and 19.44MHz. The MT optical port connector 25 is a miniaturized MT head coupler MTSX-1374-I, which is connected to the high-speed photoelectric conversion unit 22 and the multi-rate photoelectric conversion unit 23 via G653 micro-bending-resistant bare optical fiber. The function of the MT optical port connector 25 is to provide a reliable optical coupling interface for optical signal transmission between the bionic neuron optical engine module 2 and the flexible optical waveguide circuit board 1.

[0044] Furthermore, the flexible crimping unit 26 provides an electrical connection channel for the electrical signal transmission between the bionic neuron light engine module 2, the flexible light waveguide circuit board 1, and the unmanned platform; the cross-mode sensor array unit 27 provides the bionic neuron light engine module 2 with real-time sensing information including temperature, tilt angle, angular velocity, acceleration, pressure, deformation, illumination, and air pressure. In this embodiment, the flexible crimping unit 26 is a miniaturized flexible circuit board 152 contact crimp connector YRX-2673-I, which is connected to the signal processing unit 21 and the unmanned platform via PCB traces. The function of the flexible crimping unit 26 is to provide a reliable electrical connection channel for the electrical signal transmission between the bionic neuron light engine module 2, the flexible light waveguide circuit board 1, and the unmanned platform. The cross-mode sensor array unit 27 is a miniaturized low-power integrated sensor array GCXT-R-ST-221, which is connected to the domestically produced high-performance FPGA chip FM9VU13PB2104 of the signal processing unit 21 via 12 groups of LVDS buses on the flexible circuit board. The function of the cross-mode sensor array unit 27 is to provide real-time sensing information such as temperature, tilt angle, angular velocity, acceleration, pressure, deformation, illumination, and air pressure to the bionic neuron light engine module 2.

[0045] Furthermore, the edge computing CPU unit 28 is used to perform real-time edge deployment computing on the data collected and received by the bionic neuron light engine module 2; the high-speed data cache unit 29 is used to provide the signal processing unit 21 with a high-speed storage space of 128G capacity, 200GB / S throughput, and 50ns latency. In this embodiment of the invention, the edge computing CPU unit 28 is equipped with interconnected domestic control chip FMQL45T900, RSHF4644ARH power chip, and JFMFO2G16RH domestic high-performance FLASH memory chip. The domestic control chip FMQL45T900 and the domestic high-performance FLASH memory chip JFMFO2G16RH are connected together via an SPI bus. The function of RSHF4644ARH is to convert the 24V bus voltage into multiple sets of 1.0V, 1.2V, 2.5V, and 3.3V operating voltages for use by the domestic control chip FMQL45T900, the domestic high-performance FLASH memory chip JFMFO2G16RH, and the high-precision crystal oscillator OF32HB4. The edge computing CPU unit 28 is connected to the domestically produced high-performance FPGA chip FM9VU13PB2104 of the signal processing unit 21 via 24 sets of high-speed SERDES buses. The function of the edge computing CPU unit 28 is to perform real-time edge deployment computing on the data collected and received by the bionic neuron optical engine module 2, with the aim of significantly improving the real-time response time of the nodes of the "integrated sensing and computing" ultra-100G optical network neural bus. The high-speed data cache unit 29 is a high-speed, large-capacity high-speed data cache array HVES-STC-2211, which is connected to the signal processing unit 21 via 1742 parallel data and address buses. The function of the high-speed data storage unit is to provide the signal processing unit 21 with a high-speed storage space with a capacity of 128G, a throughput rate of 200GB / s, and a latency of 50ns.

[0046] Finally, the flexible optical waveguide circuit board 1 is used to provide a physical channel for the optical network neural bus to be conformally deployed on the unmanned platform for hybrid optical and electrical transmission; the power management module 3 is used to convert the 24V secondary power input from the unmanned platform into multiple power supplies of 12V, 5V, and 3.3V, and output them through the flexible optical waveguide circuit board 1 to the six bionic neuron optical engine modules 2 and the intelligent optical routing scheduling module 4 to provide power to the optical network neural bus; the intelligent optical routing scheduling module 4 is used to switch the 12 optical signals from the six bionic neuron optical engine modules 2 to the corresponding bionic neuron optical engine modules 2 according to the adaptive intelligent routing algorithm. In this embodiment of the invention, the flexible optical waveguide circuit board 1 is an ultra-thin, flexible, high-density planar fiber optic integrated circuit board RXDG-235742-I. It is connected to six bionic neuron optical engine modules 2 through a flexible electrical connection point array and an MT head, connected to a power management module 3 through a flexible electrical connection point array, and connected to an intelligent optical routing scheduling module 4 through an MT head. Its function is to provide a physical channel for optical and electrical hybrid transmission that is easy to conformally deploy on an unmanned platform for the "integrated sensing and computing" ultra-100G optical network neural bus. The power management module 3 is a miniaturized, high-power, highly integrated power module DR-24V-1664 with single-input multiple-output. It is connected to the flexible optical waveguide circuit board 1 via a flexible electrical connection point array and to the unmanned platform via a high-density power connector T9VFRED-I. Its function is to convert the 24V secondary power input from the unmanned platform into multiple power sources such as 12V, 5V, and 3.3V, and output them through the flexible optical waveguide circuit board 1 to the bionic neuron optical engine modules 21 to 26 and the intelligent optical routing scheduling module 4, providing power for the "integrated communication, sensing, and computing" ultra-100G optical network neural bus. The intelligent optical routing scheduling module 4 is a miniaturized, highly integrated, low-power 16×16 non-blocking all-optical switching matrix OXC-16-STM-I. Its function is to switch the 12 optical signals from the six bionic neuron optical engine modules 2 to the corresponding bionic neuron optical engine modules 2 according to an adaptive intelligent routing algorithm.

[0047] This invention presents a sensor-computer integrated ultra-100G optical network neural bus, developed based on a modular approach. To expand the number of bionic neuron optical engine modules 2, simply connect more modules to the flexible optical waveguide circuit board 1; the maximum number of modules can be expanded to 16. This invention provides a sensor-computer integrated ultra-100G optical network neural bus that supports seamless environmental perception, zero-latency high-bandwidth transmission, "neural reflex" level real-time response, and "cell-embedded" conformal deployment. It can meet the application requirements of various unmanned platforms such as unmanned vehicles, unmanned surface vessels, drones, and space-based satellites for multi-dimensional information perception, near-sensory real-time computing, and low-latency broadband transmission. This bus features high integration, good versatility, high intelligence, high reliability, low power consumption, low latency, low cost, easy expansion, easy iteration, and easy deployment.

[0048] Secondly, please refer to Figures 3-4 The present invention also provides a method for operating a neural bus for ultra-100G optical networks that integrates sensing, computing, and communication, comprising:

[0049] S1 Power-on Initialization: After power-on, the functional modules of the integrated inductive computing ultra-100G optical network neural bus are initialized;

[0050] In this embodiment of the invention, after power-on, the functional modules of the "integrated sensing and computing" ultra-100G optical network neural bus are initialized. The main process includes system kernel loading, register configuration, hardware enumeration and identification, system node netlist generation, driver matching and loading, driver entry initialization, PCI bus driver scanning, module subsystem registration, resource allocation, and registry control.

[0051] S2 System Self-Check: Automatically detects whether the bionic neuron light engine module 2, flexible light waveguide circuit board 1, power management module 3, and intelligent light route scheduling module 4 are working properly;

[0052] In this embodiment of the invention, the system automatically detects whether each of the bionic neuron optical engine module 2, flexible optical waveguide circuit board 1, power management module 3, and intelligent optical routing scheduling module 4 is working properly. If the register value at read address OX 0022 is OX 36, the register value at read address OX 0023 is OX 49, the register value at read address OX 0046 is OX 22, the register value at read address OX0016 is OX 11, the register value at read address OX 0087 is OX 25, and the register value at read address OX 0033 is OX... 57. If the operating voltage deviation reported by all functional modules (3.3V, 2.5V, 1.8V, 1.2V, 1.0V, 0.8V) is not less than 10%, the chip temperature is below 38℃, the HBM read / write data is correct, the CPU heartbeat counter count is correct, and the initial values ​​reported by the temperature sensor, tilt angle sensor, angular velocity sensor, acceleration sensor, pressure sensor, deformation sensor, light sensor, and air pressure sensor are correct, then the system self-test is normal.

[0053] S3 Multidimensional Sensing Data Extraction: Read the raw data collected by the cross-mode sensor array unit 27 and form raw data frames in time order. The raw data frames include image data frames, point cloud data frames, waveform data frames, and time-series numerical frames.

[0054] S4 Real-time Edge Computing: Using CPUs deployed on edge nodes, multi-dimensional sensing data extracted by the cross-mode sensor array in the bionic neuron optical engine module 2 is processed in real time.

[0055] In this embodiment of the invention, a CPU deployed on an edge node is used to perform real-time calculations on the multidimensional sensing data extracted by the cross-mode sensor array in the bionic neuron optical engine module 2. This includes Gaussian denoising, median denoising, Kalman filtering denoising, outlier removal, missing value filling, time alignment, spatial calibration, feature extraction, normalization processing, analysis and statistics, response strategy generation, and MAP table maintenance.

[0056] S5 Traffic Aggregation: The bionic neuron optical engine module 2 aggregates the large-capacity real-time data from the unmanned platform and the data generated by real-time edge computing into one data stream, which is then used to transmit the optical signal bandwidth of the flexible optical waveguide circuit board 1 to the intelligent optical routing module.

[0057] In this embodiment of the invention, the bionic neuron optical engine module 2 aggregates the large-capacity real-time data from the unmanned platform and the data generated by real-time edge computing into a single high-volume data stream, so as to fully utilize the optical signal bandwidth of the flexible optical waveguide circuit board 1 to transmit it to the intelligent optical routing module.

[0058] S6 Homomorphic Encryption: The bionic neuron light engine module 2 completes data encryption by using public key linear transformation and noise linear superposition to completely hide plaintext and ciphertext with controllable noise.

[0059] In this embodiment of the invention, in practical applications, unmanned platforms often collaborate through cloud-networked systems. This requires the "integrated sensing and computing" ultra-100G optical network neural bus to have data encryption transmission capabilities. This "integrated sensing and computing" ultra-100G optical network neural bus uses homomorphic encryption to achieve encrypted data transmission. The biomimetic neuron optical engine module 2 achieves complete plaintext hiding and controllable noise in the ciphertext through public-key linear transformation and linear noise superposition, thereby completing data encryption.

[0060] S7 error correction coding: Bionic neuron light engine module 2 adds redundant check bits to the original information code elements according to the RS coding rules;

[0061] In this embodiment of the invention, the bionic neuron light engine module 2 adds redundant check bits to the original information code elements according to the RS encoding rules to ensure that the receiving end can automatically detect transmission errors and perform automatic error correction.

[0062] S8 Intelligent Routing and Switching: Based on the MAP table provided by each bionic neuron optical engine module 2, the optimal data exchange routing matrix table of the integrated sensory computing ultra-100G optical network neural bus is generated in real time. The intelligent optical routing scheduling module 4 automatically completes the routing scheduling of all optical signals based on the data exchange routing matrix table.

[0063] S9 Error Correction Decoding: According to the RS encoding rules, the bionic neuron optical engine module 2 performs error correction decoding on the received data;

[0064] S10 Edge Trust Computing: The bionic neuron optical engine module 2 performs edge trust computing on the received broadband high-capacity data in real time to ensure that the data, code, identity, and communication are trustworthy. Data that fails the edge trust computing is isolated and the edge trust computing results are reported to other bionic neuron optical engine modules 2.

[0065] S11 Traffic Balancing: For broadband high-capacity data through edge trust computing, the signal processing unit 21 and the high-speed data cache unit 29 of the bionic neuron optical engine module 2 work together to complete traffic balancing processing, ensuring that the data is continuously transmitted to the unmanned platform at a stable rate, and avoiding data loss caused by the unmanned platform's inability to process short-term ultra-large capacity data.

[0066] This invention discloses a method for operating a 100G+ optical network neural bus integrating sensing, computing, and communication. This 100G+ optical network neural bus features high intelligence, high integration, high reliability, good versatility, scalability, replaceability, upgradeability, iterability, and low cost. The method is simple to operate and highly practical. When applied to a 100G+ optical network neural bus integrating sensing, computing, and communication, it can significantly reduce maintenance difficulty, save human resources, improve system reliability, and reduce production costs.

[0067] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A neural bus for ultra-100G optical networks integrating induction and computation, characterized in that, It includes a flexible optical waveguide circuit board, multiple biomimetic neuron optical engine modules, a power management module, and an intelligent optical routing scheduling module; Multiple biomimetic neuron light engine modules are connected to the flexible light waveguide circuit board, the power management module is connected to the flexible light waveguide circuit board, and the intelligent light routing scheduling module is connected to the flexible light waveguide circuit board.

2. The ultra-100G optical network neural bus integrating induction and computation as described in claim 1, characterized in that, The number of bionic neuron light engine modules is six. The flexible optical waveguide circuit board has a flexible electrical connection point array and an MT connector. The electrical signals of the six bionic neuron light engine modules are interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array in a press-fit manner. The optical signals of the six bionic neuron light engine modules are interconnected with the flexible optical waveguide circuit board via the MT connector in a plug-in manner. The power management module is interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array in a press-fit manner. The intelligent optical routing scheduling module is interconnected with the flexible optical waveguide circuit board via the flexible electrical connection point array in a press-fit manner.

3. The ultra-100G optical network neural bus integrating induction and computation as described in claim 1, characterized in that, The bionic neuron optical engine module includes a signal processing unit, a high-speed photoelectric conversion unit, a multi-rate photoelectric conversion unit, a clock unit, an MT optical port plug-in unit, a flexible crimp contact unit, a cross-mode sensor array unit, an edge computing CPU unit, and a high-speed data cache unit. The high-speed photoelectric conversion unit, the multi-rate photoelectric conversion unit, and the clock unit are respectively connected to the signal processing unit; the MT optical port connector unit is respectively connected to the high-speed photoelectric conversion unit and the multi-rate photoelectric conversion unit via optical fibers; the flexible crimp contact unit, the cross-mode sensor array unit, the edge computing CPU unit, and the high-speed data cache unit are respectively connected to the signal processing unit.

4. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The signal processing unit is composed of an FPGA chip of model FM9VU13PB2104, which is used to perform protocol conversion and interface matching processing on various acquired signals and data.

5. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The high-speed photoelectric conversion unit is used to complete the electro-optical and photoelectric signal conversion of one 100G or one 200G electrical signal; the multi-rate photoelectric conversion unit is used to complete the electro-optical and photoelectric signal conversion of one 155M or one 622M or one 1.25G or one 2.5G or one 10G electrical signal.

6. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The clock unit is used to provide the signal processing unit with four high-precision, low-jitter clocks at frequencies of 156.25M, 155.52M, 174M, and 19.44M; the MT optical port plug-in unit is used to provide an optical coupling interface for optical signal transmission between the bionic neuron optical engine module and the flexible optical waveguide circuit board.

7. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The flexible pressure contact unit is used to provide an electrical connection channel for the electrical signal transmission between the bionic neuron light engine module and the flexible light waveguide circuit board and the unmanned platform; the cross-mode sensor array unit is used to provide the bionic neuron light engine module with real-time sensing information including temperature, tilt angle, angular velocity, acceleration, pressure, deformation, illumination and air pressure.

8. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The edge computing CPU unit is used to perform real-time edge deployment computing on the data collected and received by the bionic neuron light engine module; the high-speed data cache unit is used to provide the signal processing unit with a high-speed storage space with a capacity of 128G, a throughput rate of 200GB / S, and a latency of 50ns.

9. The ultra-100G optical network neural bus integrating induction and computation as described in claim 3, characterized in that, The flexible optical waveguide circuit board is used to provide a physical channel for the optical network neural bus to be conformally deployed on an unmanned platform for hybrid optical and electrical transmission. The power management module is used to convert the 24V secondary power input from the unmanned platform into multiple power supplies of 12V, 5V, and 3.3V, and output them through the flexible optical waveguide circuit board to the six bionic neuron optical engine modules and the intelligent optical routing scheduling module to provide power to the optical network neural bus. The intelligent optical routing scheduling module is used to switch the 12 optical signals from the six bionic neuron optical engine modules to the corresponding bionic neuron optical engine modules according to an adaptive intelligent routing algorithm.

10. A method for operating a 100G+ optical network neural bus integrating induction and computing, applied to the 100G+ optical network neural bus integrating induction and computing as described in any one of claims 1-9, characterized in that, include: S1 Power-on Initialization: After power-on, the functional modules of the integrated inductive computing ultra-100G optical network neural bus are initialized; S2 System Self-Check: Automatically detects whether the bionic neuron light engine module, flexible light waveguide circuit board, power management module, and intelligent light route scheduling module are working properly; S3 Multidimensional Sensing Data Extraction: Reads the raw data collected by the cross-mode sensor array unit and forms raw data frames in time order. The raw data frames include image data frames, point cloud data frames, waveform data frames, and time-series numerical frames. S4 Real-time Edge Computing: Uses CPUs deployed on edge nodes to perform real-time computation on multi-dimensional sensing data extracted by the cross-mode sensor array in the bionic neuron optical engine module; S5 Traffic Aggregation: The bionic neuron optical engine module aggregates the large-capacity real-time data from the unmanned platform and the data generated by real-time edge computing into one data stream, which is then used to transmit the optical signal bandwidth of the flexible optical waveguide circuit board to the intelligent optical routing module. S6 Homomorphic Encryption: The bionic neuron light engine module completely hides the plaintext and incorporates controllable noise into the ciphertext through public key linear transformation and linear noise superposition, thus completing data encryption. S7 error correction coding: The bionic neuron light engine module adds redundant check bits to the original information code elements according to the RS coding rules; S8 Intelligent Routing and Switching: Based on the MAP table provided by each bionic neuron optical engine module, it analyzes and generates the optimal data exchange routing matrix table for the integrated sensory computing ultra-100G optical network neural bus in real time. The intelligent optical routing scheduling module automatically completes the routing scheduling of all optical signals based on the data exchange routing matrix table. S9 Error Correction Decoding: The bionic neuron optical engine module performs error correction decoding on the received data according to the RS encoding rules; S10 Edge Trust Computing: The bionic neuron optical engine module performs edge trust computing on the received broadband high-capacity data in real time to ensure that the data, code, identity, and communication are trustworthy. Data that fails the edge trust computing is isolated and the edge trust computing results are reported to other bionic neuron optical engine modules. S11 Traffic Balancing: For broadband, high-capacity data transmitted via edge trust computing, the signal processing unit and high-speed data caching unit of the bionic neuron optical engine module work together to perform traffic balancing, ensuring that data is continuously transmitted to the unmanned platform at a stable rate and avoiding data loss caused by the unmanned platform's inability to handle short-term ultra-large-capacity data.

Images