WO2021169735A1 - 一种基于深度神经网络的单发多收车灯联网系统 - Google Patents

一种基于深度神经网络的单发多收车灯联网系统 Download PDF

Info

Publication number
WO2021169735A1
WO2021169735A1 PCT/CN2021/074588 CN2021074588W WO2021169735A1 WO 2021169735 A1 WO2021169735 A1 WO 2021169735A1 CN 2021074588 W CN2021074588 W CN 2021074588W WO 2021169735 A1 WO2021169735 A1 WO 2021169735A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
dnn
networking system
equalizer
module
Prior art date
Application number
PCT/CN2021/074588
Other languages
English (en)
French (fr)
Inventor
迟楠
王超凡
Original Assignee
珠海复旦创新研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 珠海复旦创新研究院 filed Critical 珠海复旦创新研究院
Publication of WO2021169735A1 publication Critical patent/WO2021169735A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/11—Arrangements specific to free-space transmission, i.e. transmission through air or vacuum
    • H04B10/114—Indoor or close-range type systems
    • H04B10/116—Visible light communication
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/50—Transmitters
    • H04B10/501—Structural aspects
    • H04B10/502—LED transmitters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/50—Transmitters
    • H04B10/516—Details of coding or modulation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/60—Receivers
    • H04B10/66—Non-coherent receivers, e.g. using direct detection
    • H04B10/69—Electrical arrangements in the receiver
    • H04B10/697—Arrangements for reducing noise and distortion
    • H04B10/6971—Arrangements for reducing noise and distortion using equalisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H04W4/44—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for communication between vehicles and infrastructures, e.g. vehicle-to-cloud [V2C] or vehicle-to-home [V2H]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H04W4/46—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for vehicle-to-vehicle communication [V2V]

Definitions

  • the invention belongs to the technical field of visible light communication, and in particular relates to a single-transmit and multiple-receive vehicle lamp networking system based on a deep neural network.
  • the Internet of Vehicles is the concrete realization of the Internet of Things in the transportation field. It is a comprehensive network system that uses vehicles to realize the interconnection of all things and the intercommunication of information. It uses a variety of advanced technologies to obtain various information and data by establishing communication between the vehicle and the surrounding things, and analyze and make decisions based on the obtained information. It is an effective way to solve problems such as congestion and safety. It can provide a variety of services such as vehicle safety, road maintenance, traffic monitoring, life entertainment, mobile Internet access, etc., not only can significantly protect the safety of drivers, but also significantly improve the overall traffic efficiency. It can also provide richer and more considerate services for vehicles and drivers.
  • the communication system As the basis of the Internet of Vehicles, directly determines its transmission performance, and has received extensive attention.
  • the wireless network has encountered various problems and challenges in the application of the Internet of Vehicles due to some shortcomings and limitations of its own communication and transmission.
  • LED-based visible light communication is regarded as a communication method with multiple advantages, such as rich spectrum resources, energy saving, high efficiency, and green safety, which have caused a wide range of Notice.
  • the vehicle can directly use the car’s own LED lights to communicate without additional development of the transmitter, only the corresponding receiver needs to be installed, and even the vehicle’s own equipment can be converted into a receiver.
  • the complexity of development and transformation Small and low cost.
  • the research of visible light communication applied to car lights is still in its infancy, and there are fewer modulation methods studied, and because of the high power, high voltage, high heat generation of the car lights, it is difficult to drive, and the response is slow. Compared with other devices, it realizes its high-speed communication. It is a big challenge.
  • the research on visible light communication based on LED car lights is almost still in the blank stage, especially for car LED headlights.
  • visible light is used for communication, and the transmitting end is composed of various on-board units and traffic systems. That is, the vehicle itself includes headlights and taillights, other vehicles, street lights, traffic lights, LED signs, etc. These transportation systems are connected to the backbone network through a roadside infrastructure (RSI) network. VLC technology is used to transmit and broadcast data between vehicles and vehicles. For vehicle-mounted units, the headlights and taillights are usually connected to a visible light communication transceiver at the same time as the "transmitting antenna" for optical communication.
  • Photodiode (PD) is usually used as a receiver for optical communication in the Internet of Vehicles, installed on a vehicle and connected to a visible light communication transceiver.
  • image sensors imaging sensors, IS
  • PD digital signal processing circuitry
  • IS image sensors
  • Multi-source information realizes the support of rich and diverse intelligent transportation scenarios.
  • the purpose of the present invention is to provide a deep neural network (DNN)-based single-transmit and multiple-receive vehicle light networking system that can effectively improve the signal-to-noise ratio, receiving area and receiving angle for the LED car light networking system, abbreviated as SIMO -DNN car lights networking system.
  • DNN deep neural network
  • the SIMO-DNN car light networking system includes a transmitting terminal and a receiving terminal; the transmitting terminal adopts a commercial LED car light to modulate the signal to the LED for transmission; the receiving terminal includes multiple PIN arrays and transimpedance amplification modules connected in sequence , Electrical amplification module (EA), analog-to-digital converter (ADC), signal synchronization, deep neural network (DNN equalizer), maximum ratio combining module; multiple PIN arrays are used to receive multiple optical signals from the transmitter; multiple PIN arrays
  • the received multi-channel optical signals respectively pass through the transimpedance amplifier module, the electrical amplifier module (EA), the analog-to-digital converter (ADC), the signal synchronization, and the DNN equalizer. After being equalized by the DNN equalizer, each signal passes through The maximum ratio combining module performs weighted summation to achieve maximum ratio combining into a signal, and finally completes signal demodulation according to the modulation format.
  • the invention adopts a deep neural network DNN (equalizer), which can effectively suppress signal non-linearity; the signals of multiple channels are respectively passed through the DNN equalizer for signal equalization.
  • DNN deep neural network
  • Figure 3 is a fully connected network structure including an input layer, a hidden layer and an output layer.
  • the DNN equalizer that each signal passes through, the number of nodes required by each layer of the network and the number of network iterations can be debugged according to the signal characteristics to achieve the best results.
  • the multi-PIN array is composed of a plurality of PIN arrangements, which are used for receiving signals and outputting each optical signal separately; the multi-channel optical signals are respectively subjected to transimpedance amplification through the transimpedance amplification module.
  • the transimpedance amplification module and the multi-PIN array are designed on the same PCB to reduce the noise introduced by the signal during the amplification process.
  • the multiple PIN array can be expressed as an M ⁇ N arrangement, where M is 1, 2, ....
  • multiple PINs may have the same wavelength response range or different wavelength response ranges, mainly in the visible light band and the infrared band.
  • multi-PIN array if multiple PINs with different wavelength response ranges are used, they can be arranged in a variety of ways, such as a striped arrangement or a cross arrangement.
  • the plane of the multi-PIN array can be a focal plane, a cylindrical surface or a spherical surface.
  • the transimpedance amplifier module is composed of multiple single-channel transimpedance amplifiers.
  • the structure of a single-channel transimpedance amplifier is shown in Fig. 2, and includes a differential amplifier, an impedance matching module, and a transformer; the optical signal received by the PIN First, pass through a differential amplifier to output two signals with opposite polarities. The two signals pass through an impedance matching module, and finally pass through a transformer to convert the two signals into one output.
  • the signal output after transimpedance amplification is transformed into a digital signal for further processing after passing through the electrical amplification module and ADC.
  • the weight coefficient when weighting and summing, can be determined according to the signal-to-noise ratio or the amplitude ratio of the signal.
  • the signal-to-noise ratio of the signal received by a single PIN receiver in the long-distance situation is extremely low.
  • the present invention combines the multiple signals received by the multi-PIN array at the maximum ratio, which can effectively improve the signal-to-noise ratio of the system.
  • the signal output after the DNN equalizer is assigned weights according to the signal-to-noise ratio of each signal, and the weighted sum is added to achieve the maximum ratio combination of the signals.
  • the maximum ratio combined structure is shown in Figure 4. So far, the reception of the SIMO-DNN with multiple PIN arrays is completed, and its overall structure is shown in Figure 5. Finally, the combined signal is demodulated according to its modulation format.
  • the present invention Compared with the existing vehicle lamp networking solution, the present invention has the following advantages:
  • the SIMO communication system implemented by multi-PIN arrays for signal reception, and then combining multiple signals according to the signal-to-noise ratio at the maximum ratio, can effectively improve the system's signal-to-noise ratio, and thus increase the transmission rate and transmission distance of the system. It is suitable for medium and long-distance car lamp networking scenarios, and can improve the quality of car lamp networking communication when the weather is greatly affected;
  • This solution uses commercial LED car lights at the transmitting end without any modification to the transmitting end.
  • the receiving solution can be directly adapted to any car light networking system and can be demodulated for different modulation formats.
  • the invention adopts a multi-PIN receiving array to combine multiple signals with the maximum ratio, which can effectively improve the signal-to-noise ratio, increase the receiving area, and improve the stability of the system; DNN equalization can effectively suppress the nonlinear damage of the system due to the high-power characteristics of the car lights , Reduce the bit error rate.
  • the invention combines multiple PIN reception and DNN equalization, which can effectively improve the signal rate, transmission distance and system stability of the car light networking system.
  • FIG. 1 is a simplified diagram of the SIMO-DNN car lamp networking system of the present invention.
  • Figure 2 shows the structure of a single-channel transimpedance amplifier.
  • Figure 3 shows the DNN network structure.
  • Figure 4 shows the maximum ratio combined structure.
  • Figure 5 shows the structure of the SIMO-DNN car light networking system.
  • Fig. 6 is a diagram showing the relationship between the number of nodes in the input layer of the system of the present invention and the signal error rate after equalization.
  • Fig. 7 is a diagram showing the relationship between the number of hidden layer nodes in the system of the present invention and the signal error rate after equalization.
  • Fig. 8 is a diagram showing the relationship between the number of network iterations of the system of the present invention and the minimum mean square error (MSE) of the training set and the test set.
  • MSE minimum mean square error
  • Figure 9 shows the relationship between the system signal amplitude Vpp and the bit error rate and the signal frequency spectrum at different transmission distances.
  • (a) is the transmission distance of 2 meters
  • (b) is the transmission distance of 3 meters
  • (d) is the transmission distance of 5 meters.
  • Figure 10 shows the transmission rate obtained by using a single PIN receiver and a 4PIN array receiver in the system of the present invention.
  • the purpose of the present invention is to improve the overall signal-to-noise ratio of the car light networking system, and to improve the system's anti-non-linear ability, so as to obtain a higher transmission rate and a longer transmission distance.
  • the following car lights network communication scenarios provide solutions.
  • the core device of the invention is composed of a multi-PIN array, a multi-channel DNN post-equalizer and a maximum ratio combining module.
  • the emission light source is a commercial LED car light. Based on the above receiving scheme, the specific working steps of the car light networking system are as follows:
  • Step 101 modulate the signal to the commercial vehicle light for signal transmission.
  • Step 102 The signal is received by the multi-PIN array, and the signal is converted into a digital signal through transimpedance amplification, electrical amplification and ADC.
  • Step 103 Perform clock synchronization on the multiple output signals, and respectively pass the DNN post-equalizer.
  • Step 104 Perform a weighted summation of the equalized signal according to its signal-to-noise ratio to complete the maximum ratio combination.
  • Step 105 Perform signal demodulation according to the modulation format of the transmitted signal.
  • Step 101 modulate the signal to the commercial vehicle light for signal transmission.
  • the signal is modulated according to actual needs, and commercial vehicle lights are used for signal transmission.
  • the modulation format adopted by this system is: DMT-bitloading.
  • Step 102 The signal is received by the multi-PIN array, and the signal is converted into a digital signal through transimpedance amplification, electrical amplification and ADC.
  • a PIN array composed of multiple PINs is used to receive signals and output each signal separately.
  • Multi-path optical signals are respectively amplified by transimpedance.
  • the transimpedance amplification module and the multi-PIN array are designed on the same PCB to reduce the noise introduced by the signal during the amplification process.
  • the transimpedance amplifying module first passes the optical signal received by the PIN to output two signals with opposite polarities through a differential amplifier, then passes the two signals through the impedance matching module, and finally converts the two signals into one output through a transformer.
  • the output signal after transimpedance amplification is converted into a digital signal for further processing after electrical amplification and ADC.
  • the multi-PIN array used in this system is a 2 ⁇ 2 4PIN array.
  • Step 103 Perform clock synchronization on the multiple output signals, and respectively pass the DNN post-equalizer.
  • the multi-channel signals are respectively passed through the DNN equalizer for signal equalization.
  • the DNN equalizer network structure adopted by this system includes a fully connected network structure of an input layer, a hidden layer and an output layer.
  • Figure 6- Figure 8 shows the process of adjusting the DNN network structure.
  • Figure 6 is the relationship between the number of input layer nodes and the signal error rate after equalization
  • Figure 7 is the relationship between the number of hidden layer nodes and the signal error rate after equalization.
  • Figure 8 shows the relationship between the number of network iterations and the minimum mean square error (MSE) of the training set and the test set. It can be seen that through DNN training, the MSE of the test set decreases, and with the increase of epoch, it tends to be saturated, and there is no under-fitting and over-fitting. The fitting phenomenon proves the effectiveness and robustness of the neural network.
  • Figure 9 is a diagram showing the relationship between the system signal amplitude Vpp and the bit error rate.
  • the system bit error rate through the DNN equalizer is lower than the original signal, and as the signal amplitude Vpp increases, the optimization effect of the DNN equalizer is more obvious.
  • the DNN equalizer can effectively resist non-linearity and is suitable for high-power automotive lamp networking systems.
  • Step 104 Perform a weighted summation of the equalized signal according to its signal-to-noise ratio to complete the maximum ratio combination.
  • FIG. 10 shows the transmission rate of the system using a single PIN receiver and a 4PIN array receiver. It can be seen that multiple PIN receiving arrays can effectively increase the system transmission rate under the same conditions.
  • Step 105 Perform signal demodulation according to the modulation format of the transmitted signal.
  • the corresponding signal demodulation is performed according to the modulation format of the transmitted signal to complete the signal transmission process.
  • the complete SIMO-DNN receiving method for car lights networking has been completed.
  • Scope Examples such as changing the number of PINs in a multi-PIN array, or changing the post-DNN equalizer network structure and the number of nodes in each layer, the number of iterations, or changing the weight of each signal in the maximum ratio combination, etc., do not deviate from the spirit and the spirit of the present invention. Scope.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Electromagnetism (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Dc Digital Transmission (AREA)
  • Optical Communication System (AREA)

Abstract

本发明属于可见光通信技术领域，具体为一种基于深度神经网络的单发多收车灯联网系统。本发明系统包括发射端和接收端；发射端采用LED车灯，将信号调制至LED上进行发射；接收端采用多PIN阵列，用于接收发射端发出的多路光信号；接收到的多路光信号分别依次经过跨阻放大模块、电放大模块、模数转换器、信号同步、DNN均衡器；由DNN均衡器进行均衡后，各路信号再经过最大比合并模块，进行加权求和，实现最大比合并为一路信号，最后根据调制格式完成信号解调；本发明将多PIN接收与DNN均衡相结合，可有效提升车灯联网系统的信号速率、传输距离和系统稳定性。

Description

一种基于深度神经网络的单发多收车灯联网系统 技术领域
本发明属于可见光通信技术领域，具体涉及一种基于深度神经网络的单发多收车灯联网系统。
背景技术
车联网(IoV)是物联网在交通领域的具体实现，是利用车辆实现万物互联、信息互通的综合网络系统。它利用了多种先进的技术，通过建立车辆与周遭事物之间的通信，获取各类信息与数据，并根据获取到的信息进行分析与决策，是解决拥堵、安全等问题的有效途径。它可以提供车载安全、道路养护、交通监控、生活娱乐、移动互联网接入等丰富多彩的服务，不仅能够对驾驶员的安全起到显著的保护作用，对整体交通效率起到明显的提升作用，还能够为车辆及驾驶人员提供更丰富、更贴心的服务。而随着车联网的持续发展，通信系统作为车联网的基础，直接决定其传输性能，得到了广泛的关注。随着业务场景的不断丰富，无线网络因其本身通信传输的一些短板和限制，在车联网的应用中遇到了种种问题和挑战。与此同时，随着LED在车灯及交通信号灯制造中的广泛使用，基于LED的可见光通信被视为具备多种优势的通信方式，如频谱资源丰富、节能高效、绿色安全，引起了广泛的注意。除此之外，车辆可以直接利用汽车自带的LED车灯进行通信，无需额外开发发射端，只需要安装对应的接收机，甚至可以利用车辆本身的设备改装成接收机，开发、改造复杂度小，成本低。但是应用于车灯的可见光通信研究仍处于起步阶段，研究的调制方式较少，且由于车灯功率大，电压高，发热大，难以驱动，响应慢，与其他器件相比，实现其高速通信是一个很大的挑战。而在基于LED车灯的可见光通信这方面的研究几乎还是处在空白阶段，特别是汽车LED前照灯。
车联网中利用可见光进行通信，发射端由各种车载单元和交通系统组成，即车辆本身包括前照大灯和汽车尾灯、其他车辆、路灯、交通灯、LED标牌等。这些交通系统通过路边基础设施(RSI)网络连接到主干网络，车与车、车与设备之间通过VLC技术来进行数据的传输和广播。对于车载单元来说，通常将前照灯和尾灯同时连接可见光通信收发器，作为光通信的“发射天线”。光电二极管(Photodiode，PD)通常作为光通信在车联网中的接收机，安装在车辆上并连接至可见光通信收发器。此外也可以利用摄像头等图像传感器(imaging sensor,IS)或者PD和IS的结合来作为通信的接收端，方便在接收端对传输的信号进行解调，从而获取车辆本身的信息、道路信息及更多来源的信息，实现丰富多样智能交 通场景支撑。
发明内容
本发明的目的在于为基于LED车灯联网系统提供一种可有效提高信号信噪比、接收面积与接收角度的基于深度神经网络(DNN)的单发多收车灯联网系统，简记为SIMO-DNN车灯联网系统。
本发明提供的SIMO-DNN车灯联网系统，包括发射端和接收端；发射端采用商用LED车灯，将信号调制至LED上进行发射；接收端包括依次连接的多PIN阵列、跨阻放大模块、电放大模块(EA)、模数转换器(ADC)、信号同步、深度神经网络即DNN均衡器、最大比合并模块；多PIN阵列用于接收发射端发出的多路光信号；多PIN阵列接收到的多路光信号分别依次经过跨阻放大模块、电放大模块(EA)、模数转换器(ADC)、信号同步、DNN均衡器；由DNN均衡器进行均衡后，各路信号再经过最大比合并模块，进行加权求和，实现最大比合并为一路信号，最后根据调制格式完成信号解调。
由于车灯发射功率大，导致光信号受非线性影响严重，需经过均衡器以提升系统性能。本发明采用深度神经网络DNN(均衡器)，可有效压制信号非线性；将多路信号分别经过DNN均衡器进行信号均衡。DNN均衡器结构如图3所示，是包含一个输入层、一个隐藏层和一个输出层的全连接网络结构。各路信号经过的DNN均衡器，其网络各层所需的节点数以及网络迭代次数，可根据信号特征进行调试，以达到最佳效果。
本发明中，多PIN阵列由多个PIN排列组成，用于接收信号并将各路光信号分别输出；多路光信号分别经过跨阻放大模块进行跨阻放大。跨阻放大模块与多PIN阵列设计在同一块PCB上以降低信号在放大过程中引入的噪声。
本发明中，所述多PIN阵列，可表示为M×N的排列，M为1，2，…。
所述多PIN阵列中，多个PIN可以具有相同的波长响应范围，亦可具有不同的波长响应范围，主要为可见光波段及红外波段。
所述多PIN阵列中，若采用不同波长响应范围的多个PIN，可采用多种方式进行排列，条形排列或者交叉排列。
所述多PIN阵列，其平面可以是焦平面、柱面或者是球面。
本发明中，所述跨阻放大模块由多个单路跨阻放大器组成，个单路跨阻放大器结构如图2所示，包括一个差分放大器、阻抗匹配模块、变压器；PIN接收到的光信号首先经过差分放大器，输出两路极性相反的信号，两路信号经过阻抗匹配模块，最后经过变压器，将两路信号转变为一路输出。经过跨阻放大输出的信号再经过电放大模块和ADC后转变为数 字信号供下一步处理。
最大比合并模块中，加权求和时，其权重系数可根据信号的信噪比比例或幅值比例确定。
单个PIN的接收机在远距离情况下收到的信号信噪比极低，本发明将多PIN阵列接收到的多个信号进行最大比合并，可有效提升系统信噪比。将经过DNN均衡器后输出的信号，根据各路信号的信噪比分配权重，加权求和，以实现信号的最大比合并。最大比合并结构如图4所示。至此，多PIN阵列的SIMO-DNN的接收完成，其整体结构如图5所示。最后再将合并后的信号再根据其调制格式进行解调。
本发明与现有的车灯联网方案相比，具有以下优越性：
(1)通过多PIN阵列进行信号接收，再对多路信号根据信噪比进行最大比合并实现的SIMO通信系统，可有效提升系统信噪比，并可因此提升系统的传输速率、传输距离，适用于中远距离的车灯联网场景，并可一定程度上提升受天气影响较大时的车灯联网通信质量；
(2)通过DNN后均衡器对接收信号进行均衡，可有效抗信号非线性所带来的不利影响。而现在的车灯联网系统其工作功率相对较大，信号受非线性影响极大，因此本发明适用于现有的车灯通信方案；
(3)本方案在发射端采用商用LED车灯，未对发射端进行任何改造，该接收方案可直接适配于任何车灯联网系统，并且可针对不同调制格式进行解调。
本发明采用多PIN接收阵列通过多路信号最大比合并，可有效提升信噪比，增大接收面积，提升系统稳定性；DNN均衡可有效抑制系统由于车灯高功率特性带来的非线性损伤，降低误码率。本发明将多PIN接收与DNN均衡相结合，可有效提升车灯联网系统的信号速率、传输距离和系统稳定性。
附图说明
图1为本发明SIMO-DNN车灯联网系统简图。
图2为单路跨阻放大器结构。
图3为DNN网络结构。
图4为最大比合并结构。
图5为SIMO-DNN车灯联网系统结构图示。
图6为本发明本系统输入层节点数与均衡后信号误码率关系图。
图7为本发明系统隐藏层节点数与均衡后信号误码率关系图。
图8为本发明系统网络迭代次数与训练集、测试集最小均方误差(MSE)关系图。
图9为不同传输距离时系统信号幅度Vpp与误码率关系及信号频谱图。其中，(a)为传输距离为2米的，(b)为传输距离为3米的，为传输距离为4米的，(d)为输距离为5的。
图10为本发明系统分别采用单PIN接收机和4PIN阵列接收机所得的传输速率。
具体实施方式
为使本发明的目的、技术方案和优点更加清楚，下面将结合附图对本发明的各实施方式进行详细的阐述。然而，本领域的普通技术人员可以理解，在本发明各实施方式中，为了使读者更好地理解本申请而提出了许多技术细节。但是，即使没有这些技术细节和基于以下各实施方式的种种变化和修改，也可以实现本申请各权利要求所要求保护的技术方案。
本发明的目的在于将车灯联网系统的整体信噪比提高，并提升系统的抗非线性能力，以获取更高的传输速率和更远的传输距离，并为在受天气影响较大的情况下的车灯联网通信场景提供解决方案。
本发明核心装置由多PIN阵列、多路DNN后均衡器和最大比合并模块组成。发射光源为商用LED车灯。基于上述接收方案，该车灯联网系统的具体工作步骤如下：
步骤101：将信号调制到商用车灯上进行信号发射。
步骤102：多PIN阵列对信号进行接收，经过跨阻放大、电放大和ADC，将信号转变为数字信号。
步骤103：将输出的多路信号进行时钟同步，分别经过DNN后均衡器。
步骤104：均衡后的信号根据其信噪比进行加权求和，完成最大比合并。
步骤105：根据发射信号的调制格式进行信号解调。
接下来按照步骤进行详细介绍：
步骤101：将信号调制到商用车灯上进行信号发射。
根据实际需求对信号进行调制，使用商业车灯进行信号发射。本系统采用的调制格式为：DMT-bitloading。
步骤102：多PIN阵列对信号进行接收，经过跨阻放大、电放大和ADC，将信号转变为数字信号。
多个PIN组成的PIN阵列，用于接收信号并将各路信号分别输出。多路光信号分别进行跨阻放大。跨阻放大模块与多PIN阵列设计在同一块PCB上以降低信号在放大过程中引入的噪声。跨阻放大模块将PIN接收到的光信号首先经过差分放大器输出两路极性相反的信号，再将两路信号经过阻抗匹配模块，最后经过变压器将两路信号转变为一路输出。经过 跨阻放大输出的信号在经过电放大和ADC后转变为数字信号供下一步处理。本系统采用的多PIN阵列为2×2的4PIN阵列。
步骤103：将输出的多路信号进行时钟同步，分别经过DNN后均衡器。
将多路信号分别经过DNN均衡器进行信号均衡。本系统采用的DNN均衡器网络结构包含一个输入层，一个隐藏层和一个输出层的全连接网络结构。各路信号经过的DNN均衡器，其网络各层所需的节点数以及网络迭代次数，需根据信号特征进行调试以达到最佳效果。图6-图8给出了DNN网络结构调整的过程，其中，图6为输入层节点数与均衡后信号误码率关系图，图7为隐藏层节点数与均衡后信号误码率关系图，可见不同信噪比情况下需根据情况挑选相应的网络结构。图8为网络迭代次数与训练集、测试集最小均方误差(MSE)关系图，可见通过DNN训练，测试集的MSE下降，并且随着epoch的增加，趋于饱和，没有欠拟合与过拟合的现象，证明了该神经网络的有效性与鲁棒性。图9为本系统信号幅度Vpp与误码率关系图，可见通过DNN均衡器的系统误码率均比原始信号低，并且随着信号幅度Vpp的提高，DNN均衡器的优化效果越明显，可见DNN均衡器可有效抗非线性，适用于大功率的车灯联网系统中。
步骤104：均衡后的信号根据其信噪比进行加权求和，完成最大比合并。
将经过DNN均衡器后输出的信号需根据各路信号的信噪比分配权重，加权求和，以实现信号的最大比合并。图10为本系统分别采用单PIN接收机和4PIN阵列接收机所得的传输速率，可见多PIN接收阵列可有效提升相同条件下的系统传输速率。
步骤105：根据发射信号的调制格式进行信号解调。
根据实际应用情况，根据发射信号的调制格式进行相应的信号解调，完成信号的传输过程。至此，完整的车灯联网SIMO-DNN接收方法已全部完成。
本领域的普通技术人员可以理解，上述各实施方式是实现本发明的具体实施例，而在实际应用中，可以在形式上和细节上对其作各种改变，而不偏离本发明的精神和范围。示例如改变多PIN阵列中PIN的个数，或改变DNN后均衡器网络结构及各层节点数、迭代次数，抑或是改变最大比合并中各路信号权重等，都不偏离本发明的精神和范围。

Claims (7)

  1. 一种SIMO-DNN车灯联网系统，其特征在于，包括发射端和接收端；发射端采用LED车灯，将信号调制至LED上进行发射；接收端包括依次连接的多PIN阵列、跨阻放大模块、电放大模块、模数转换器、信号同步、深度神经网络即DNN均衡器、最大比合并模块；多PIN阵列用于接收发射端发出的多路光信号；多PIN阵列接收到的多路光信号分别依次经过跨阻放大模块、电放大模块、模数转换器、信号同步、DNN均衡器；由DNN均衡器进行均衡后，各路信号再经过最大比合并模块，进行加权求和，实现最大比合并为一路信号，最后根据调制格式完成信号解调；
    其中，DNN均衡器是包含一个输入层、一个隐藏层和一个输出层的全连接网络结构。
  2. 根据权利要求1所述的SIMO-DNN车灯联网系统，其特征在于，所述多PIN阵列由多个PIN排列组成，可表示为M×N的排列，M为1，2，…。
  3. 根据权利要求2所述的SIMO-DNN车灯联网系统，其特征在于，所述多PIN阵列中，多个PIN具有相同的波长响应范围，或者具有不同的波长响应范围，主要为可见光波段及红外波段。
  4. 根据权利要求3所述的SIMO-DNN车灯联网系统，其特征在于，所述多PIN阵列中，采用不同波长响应范围的多个PIN时，采用多种方式进行排列，包括条形排列或者交叉排列。
  5. 根据权利要求3所述的SIMO-DNN车灯联网系统，其特征在于，所述多PIN阵列，其平面是焦平面、柱面或者是球面。
  6. 根据权利要求1-5之一所述的SIMO-DNN车灯联网系统，其特征在于，所述跨阻放大模块由多个单路跨阻放大器组成，个单路跨阻放大器结构包括一个差分放大器、阻抗匹配模块、变压器；PIN接收到的光信号首先经过差分放大器，差分放大器输出两路极性相反的信号；两路信号经过阻抗匹配模块，最后经过变压器，将两路信号转变为一路输出；经过跨阻放大输出的信号再经过电放大模块和ADC后转变为数字信号。
  7. 根据权利要求1-5之一所述的SIMO-DNN车灯联网系统，其特征在于，最大比合并模块中，加权求和时，其权重系数可根据信号的信噪比比例或幅值比例确定。
PCT/CN2021/074588 2020-02-29 2021-02-01 一种基于深度神经网络的单发多收车灯联网系统 WO2021169735A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010132377.8A CN111355531A (zh) 2020-02-29 2020-02-29 一种基于深度神经网络的单发多收车灯联网系统
CN202010132377.8 2020-02-29

Publications (1)

Publication Number Publication Date
WO2021169735A1 true WO2021169735A1 (zh) 2021-09-02

Family

ID=71194205

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/074588 WO2021169735A1 (zh) 2020-02-29 2021-02-01 一种基于深度神经网络的单发多收车灯联网系统

Country Status (2)

Country Link
CN (1) CN111355531A (zh)
WO (1) WO2021169735A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111355531A (zh) * 2020-02-29 2020-06-30 珠海复旦创新研究院 一种基于深度神经网络的单发多收车灯联网系统

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090011431A (ko) * 2007-07-26 2009-02-02 삼성전자주식회사 가시광 통신 시스템에서 수광소자의 이득치 결정 방법 및장치
CN105099553A (zh) * 2015-08-31 2015-11-25 华南理工大学 一种基于神经元网络的可见光通信接收方法及其系统
CN105281833A (zh) * 2014-06-13 2016-01-27 南京复实通讯科技有限公司 可见光信号传输方法及其系统
KR101613989B1 (ko) * 2014-12-17 2016-04-21 울산과학기술원 Wave 통신 기반의 차량 내에서 led 통신을 이용한 데이터 패킷 처리 방법 및 그 장치
JP2017085230A (ja) * 2015-10-23 2017-05-18 株式会社デンソー 可視光通信装置
CN106921470A (zh) * 2017-04-27 2017-07-04 华南理工大学 一种基于神经元网络的可见光通信接收方法及其接收系统
CN107342814A (zh) * 2017-07-21 2017-11-10 华南理工大学 一种基于可见光通信的神经网络均衡器
KR20180065142A (ko) * 2016-12-07 2018-06-18 서울과학기술대학교 산학협력단 LED 신호등의 칼라광정보를 통한 신호등객체주변기기로의 IoT 연계형 스마트 가시광 통신장치 및 방법
CN108923856A (zh) * 2018-07-06 2018-11-30 西安电子科技大学 基于车载可见光通信的发射端led车灯自动控制装置
CN110830112A (zh) * 2019-10-16 2020-02-21 青岛海信电器股份有限公司 一种可见光通信的方法及设备
CN111355531A (zh) * 2020-02-29 2020-06-30 珠海复旦创新研究院 一种基于深度神经网络的单发多收车灯联网系统

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108631867A (zh) * 2018-01-05 2018-10-09 常州星宇车灯股份有限公司 一种基于led车灯的车对车通信收发器及编解码方法
CN110091797A (zh) * 2019-05-10 2019-08-06 北京工业大学 一种基于刹车强度与转向角度信息的车灯频闪通信系统

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090011431A (ko) * 2007-07-26 2009-02-02 삼성전자주식회사 가시광 통신 시스템에서 수광소자의 이득치 결정 방법 및장치
CN105281833A (zh) * 2014-06-13 2016-01-27 南京复实通讯科技有限公司 可见光信号传输方法及其系统
KR101613989B1 (ko) * 2014-12-17 2016-04-21 울산과학기술원 Wave 통신 기반의 차량 내에서 led 통신을 이용한 데이터 패킷 처리 방법 및 그 장치
CN105099553A (zh) * 2015-08-31 2015-11-25 华南理工大学 一种基于神经元网络的可见光通信接收方法及其系统
JP2017085230A (ja) * 2015-10-23 2017-05-18 株式会社デンソー 可視光通信装置
KR20180065142A (ko) * 2016-12-07 2018-06-18 서울과학기술대학교 산학협력단 LED 신호등의 칼라광정보를 통한 신호등객체주변기기로의 IoT 연계형 스마트 가시광 통신장치 및 방법
CN106921470A (zh) * 2017-04-27 2017-07-04 华南理工大学 一种基于神经元网络的可见光通信接收方法及其接收系统
CN107342814A (zh) * 2017-07-21 2017-11-10 华南理工大学 一种基于可见光通信的神经网络均衡器
CN108923856A (zh) * 2018-07-06 2018-11-30 西安电子科技大学 基于车载可见光通信的发射端led车灯自动控制装置
CN110830112A (zh) * 2019-10-16 2020-02-21 青岛海信电器股份有限公司 一种可见光通信的方法及设备
CN111355531A (zh) * 2020-02-29 2020-06-30 珠海复旦创新研究院 一种基于深度神经网络的单发多收车灯联网系统

Also Published As

Publication number Publication date
CN111355531A (zh) 2020-06-30

Similar Documents

Publication Publication Date Title
CN101282175B (zh) 基于垂直分层空时编码的自由空间mimo光通信系统
CN107040310B (zh) 基于ihdaf协议的fso通信系统的工作方法
CN103117796A (zh) 卫星射频与光纤信号传输数据的中继转发装置及转发方法
CN105099553B (zh) 一种基于神经元网络的可见光通信接收方法及其系统
CN104539738A (zh) 一种led通信的车辆自组织网络
WO2021169735A1 (zh) 一种基于深度神经网络的单发多收车灯联网系统
CN107210814A (zh) 用于优化vlc型双向传输流量的与光伏模块相关的电子设备
CN110557201A (zh) 一种水下可见光通信移动节点的多角度接收装置
CN104253647A (zh) 基于mimo技术和ofdm技术的室外可见光通信系统
CN103051375B (zh) 无线激光通信外差检测系统及其检测方法
WO2021203487A1 (zh) 光纤使能光无线通信系统及方法
CN106375004A (zh) 一种基于哈特莱变换的可见光通信空间调制方法及其实现系统
CN101540635B (zh) Led交通灯可见光无线通信接收机的接收方法及应用方法
Haas et al. An introduction to optical wireless mobile communication
CN110995347A (zh) 一种适合于道路交通安全中使用的led可见光通信系统
Yahia et al. Performance evaluation of vehicular Visible Light Communication based on angle-oriented receiver
CN102158281A (zh) 高速全向无线激光通信设备
CN208986943U (zh) 自由空间光通信串并行组合中继装置
CN115333634B (zh) 一种适用于商用照明led的高带宽、多自由度无线光通信系统
Albayati An overview of visible light communication systems
CN114189282B (zh) 一种无人机辅助的无线光通信系统传输方法
CN113285763A (zh) 一种蓝绿led通信中分集接收的方法
Wang et al. Experimental testing of high-capacity bandwidth efficient visible light communication with silicon-based RGBY-LED
O'Brien Indoor optical wireless communications: recent developments and future challenges
CN108494483B (zh) 一种公路光无线通信方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21760174

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21760174

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16/02/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21760174

Country of ref document: EP

Kind code of ref document: A1