CN204948075U - A kind of visible light communication receiving system based on neuroid - Google Patents

Abstract

The utility model discloses a neuron network-based visible light communication receiving method, which comprises the following steps: 1. A first programmable gate array converts a video signal sent by a signal source into a digital signal; 2. The digital signal is driven by an LED. The circuit drives the LED lamp to emit light signals; 3. The light signals enter the receiving subsystem through n photoelectric detection devices; 4. The n data streams are input to the trained neuron merging module; 5. The The neuron merging module equalizes the n data streams and outputs the combined data stream; the combined data stream is converted into a video signal through the second programmable gate array demodulation process. The utility model also discloses a visible light communication receiving system for implementing the neuron network-based visible light communication receiving method, including: a transmitting subsystem, a transmitting subsystem and a receiving subsystem. The utility model has the advantages of wide application prospect and the like.

Description

A visible light communication receiving system based on neuron network

technical field

The utility model relates to visible light communication technology, in particular to a visible light communication receiving system based on neuron network.

Background technique

Compared with traditional infrared and wireless communication, LED visible light communication technology has the advantages of high transmission power, no electromagnetic interference, no need to apply for spectrum resources and confidentiality of information. However, visible light communication still faces many challenges, one of the biggest challenges is that inter-symbol interference greatly limits the data transmission rate of visible light communication systems. White LEDs used for lighting are generally in the form of LED arrays. Different point light source LEDs correspond to different optical paths, and the delay in signal transmission between optical paths will cause inter-symbol interference; at the same time, when the system data transmission rate is relatively high, due to the LED The limitation of bandwidth will cause the influence of one signal to spread to adjacent signals, resulting in intersymbol interference and greatly increasing the bit error rate of the system; and due to the unsatisfactory channel, symbols will be widened and delayed during transmission. The waveform shows that the symbol pulses produce tails, and the tails of adjacent pulses will overlap each other, which will also cause inter-symbol interference, which increases the bit error rate and affects the quality of communication. In general, intersymbol interference can be reduced by changing the coding method, such as changing the non-return-to-zero code in the OOK modulation method to a return-to-zero code or using orthogonal frequency division multiplexing technology. However, it greatly increases the complexity of the visible light communication system.

In communication fields such as scattering, relay, microwave, etc., diversity reception technology is often used to solve the multipath effect caused by the instability of channel parameters. Diversity reception technology is that the same signal is dispersedly transmitted through different paths, time, angles, frequencies, etc., and the receiving end obtains multiple sets of independent signals, and through appropriate combination methods, such as selective combination, equal gain combination or maximum ratio addition, etc.; Combine multiple sets of independent signals into a total received signal. At the same time, in the diversity receiving system, since the receiving module has multiple photodetectors, it is equivalent to increasing the receiving area of the photodiode, thereby providing diversity gain for the system, so the diversity receiving technology can be used to improve the performance of the visible light communication system. However, due to the time-varying and random nature of the wireless channel, different combining methods are selected in different environments; and the received data is the original data disturbed in the channel, and the data is not further optimized.

Utility model content

In order to overcome the above-mentioned shortcomings and deficiencies of the prior art, the primary purpose of this utility model is to provide a visible light communication receiving method based on a neuron network. The visible light communication receiving method further optimizes the channel performance of the VLC system without increasing the bandwidth of the device. Under this circumstance, the quality and data transmission rate of wireless communication are doubled.

In order to overcome the above-mentioned shortcomings and deficiencies of the prior art, another object of the present utility model is to provide a visible light communication receiving system based on a neural network-based visible light communication receiving method. The visible light communication receiving system is based on a simple baseband modulation technology. Diversity reception technology is used to weaken the impact of intersymbol interference on the system, and the artificial neural network is used to combine and optimize the data obtained by diversity reception to reduce the bit error rate of the system. It overcomes the diversity of combining methods of traditional diversity receiving technology in different environments.

The primary purpose of the utility model is achieved through the following technical solutions: a neuron network-based visible light communication receiving method, comprising the following steps:

Step 1, the first programmable gate array converts the video signal sent by the source into a digital signal;

Step 2, the digital signal drives the LED lamp to emit a light signal through the LED drive circuit;

Step 3, the optical signal enters the receiving subsystem through n photoelectric detection devices, wherein n is a positive integer; the n photodetectors correspond to n sub-channels of the transmission subsystem; the n sub-channels correspond to the receiving subsystem of n data streams;

Step 4, after the n data streams are amplified and filtered, they are input to the trained neuron merging module;

Step 5. The neuron merging module equalizes the n data streams and outputs a combined data stream; the combined data stream is converted into a video signal through demodulation by the second programmable gate array.

The trained neuron merging module in the step 5 includes the following steps:

Step 41. During training, give a set of input values and matching expected values to the neuron network, adjust the connection weights according to this set of training data, and pass the expected values and the actual output values of forward propagation Compare to get the error signal;

${\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>$

E(x)=[y(x)-F(x)] ² ;

Step 42, using the gradient descent method to perform error backpropagation and weight correction, and make the error signal reach or be lower than the set value through repeated learning;

${\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}\frac{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{<span\; class="notranslate">\; \&part;</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; ;</span>$

Step 43, through the training of the above equation, the weighted value of the connection can be adjusted to the minimum mean square error between the actual output of the neural network and the expected output; after training with the BP algorithm, for any input value, the neuron All equalizers can give a relatively suitable output to complete the channel equalization process.

Another object of the utility model can be achieved through the following technical solutions: a visible light communication receiving system for implementing the neuron network-based visible light communication receiving method, including: a transmitting subsystem, a transmitting subsystem and a receiving subsystem, the The transmitting subsystem has: a first programmable gate array, LED drive circuit and LED lamps, and the receiving subsystem has: n photodetection devices, ADC analog-to-digital converter, neuron merging module and a second programmable gate array The transmission subsystem is used to transmit the visible light signal sent by the LED lamp to n photodetection devices; the n photodetection devices correspond to the n sub-channels of the transmission subsystem; the first programmable gate array, LED drive Circuits, LED lamps, photoelectric detection devices, ADC analog-to-digital converters, neuron merging modules, and the second programmable gate array are connected in sequence; the first programmable gate array converts video signals into digital signals, and converts the digital The signal is transmitted to the LED drive circuit; the LED drive circuit drives the LED lamp to emit light signals; the light signal enters the photoelectric detection device through free space; the n photoelectric detection devices convert the light signal into an analog signal to form n channels of data stream; the ADC digital-to-analog converter samples the analog signal output by the photoelectric detection device into n-way digital signals; the n-way digital signals are combined and equalized by the neuron merging module, and then input to the second programmable gate array; the second programmable gate array restores the signal to a video signal after demodulation.

The LED driving circuit includes: a signal source, a variable resistor, a high-speed buffer, a BiasTee module, a DC current source and a current limiting resistor, and the described signal source, a variable resistor, a high-speed buffer, a BiasTee module and a current limiting resistor are sequentially connected; the positive pole of the DC current source is connected to the signal source, and the negative pole of the DC current source is connected to the BiasTee module; the BiasTee module includes a capacitor and an inductor; one end of the inductor is connected to the negative pole of the DC current source, and the The other end of the inductance is connected to the negative pole of the capacitor, and the positive pole of the capacitor is connected to the high-speed buffer; the electrical signal output by the source is transmitted to the BiasTee module through the high-speed buffer, and the direct current signal output by the direct current source and the high-speed buffer The signal transmitted by the high-speed buffer is coupled in the BiasTee module to generate a coupled electrical signal; the coupled electrical signal is output to the LED lamp through a current limiting resistor.

The merging processing of the neuron merging module is completed by a programmable gate array signal processing chip or a digital signal processing chip.

The transmitting subsystem also includes: a first liquid crystal display and a camera, and the receiving subsystem also includes an amplification circuit, a filter circuit and a second liquid crystal display; the first liquid crystal display and the camera are all connected to the first programmable gate array connection, the photodetection device is connected with the ADC digital-to-analog converter through the pre-amplification circuit and the post-amplification circuit, and the second programmable gate array is connected with the second liquid crystal display; the video camera transmits the video signal to the first programmable gate Array; the first programmable gate array transmits the video signal to the first liquid crystal display; the photoelectric detection device amplifies the analog signal through the amplifying circuit; the amplified analog signal is filtered and then transmitted through the filter circuit To the ADC analog-to-digital converter; the second programmable gate array transmits the video signal to the second liquid crystal display.

Another object of the utility model can also be achieved through the following technical solutions: a visible light communication receiving system based on a neural network-based visible light communication receiving method, including: a transmitting subsystem, a transmitting subsystem and a receiving subsystem, the transmitting subsystem The system inputs data into the LED lamps through the LED drive circuit, emits visible light, effectively transmits the optical signal through the transmission subsystem, and finally realizes the communication between the change of the sensed light intensity and the conversion of data through the receiving subsystem.

The transmitting subsystem is composed of a camera, a first programmable gate array, a first liquid crystal display, an LED driving circuit, and an LED lamp. Further, the LED driving circuit is composed of electrical components such as capacitors and inductors.

The transmission subsystem is free space and has n sub-channels, where n is a positive integer.

The receiving subsystem is composed of n photodetectors, an amplifier circuit, a filter circuit, a neuron merging module, a second programmable gate array, and a second liquid crystal display.

The principle of the utility model: the utility model utilizes the diversity receiving technology and the neuron merged network. The diversity receiving technology is that the same signal is dispersedly transmitted through different paths, time, angles, frequencies, etc., and multiple photoelectric The detection device obtains multiple sets of independent signals, and through appropriate combining methods, such as selective combination, equal gain combination, or maximum ratio addition, etc.; multiple sets of independent signals are combined into a total received signal. At the same time, since the diversity reception technology has multiple photodetectors, it is equivalent to increasing the effective area of the receiver for light reception, so the signal-to-noise ratio of the signal source is increased while the signal-to-noise ratio of the source is unchanged, so that the system Lower bit error rate and stronger fault tolerance. The neuron merging module combines and optimizes the obtained multiple sets of data to form a total output signal. The nonlinear mapping relationship between input and output is realized by using a nonlinear transfer function. Then filter the interfered signal, estimate and compensate the phase shift of the channel, etc. to achieve equalization. Due to the influence of intersymbol interference, the bit error rate of the visible light communication system is increased. Therefore, the influence of intersymbol interference on the system is weakened by using diversity reception technology, and the data obtained by diversity reception is processed by artificial neural network. Combined and optimized to reduce the bit error rate of the system. The receiving system can effectively reduce the influence of intersymbol interference on the system, improve the signal-to-noise ratio of the received signal, and reduce the bit error rate of the system. Under the premise of not increasing the bandwidth of the device, the quality and data transmission rate of wireless communication can be doubled; it has broad application prospects.

Compared with the prior art, the utility model has the following advantages and beneficial effects:

1. The utility model is based on a simple baseband modulation technology, which greatly simplifies the complexity of the system compared with the traditional use of OFDM modulation and discrete multi-tone modulation technology, and can be achieved without increasing the bandwidth of the LED device. , which doubles the quality of wireless communication and the capacity of the channel.

2. The utility model uses the artificial neuron network to combine and optimize multiple sets of data obtained by the diversity receiving technology. While eliminating the influence of inter-symbol interference, the utility model reduces the bit error rate and plays the role of post-equalization.

Description of drawings

FIG. 1 is a schematic diagram of the present invention implementing a VLC receiving system based on a neuron network.

Fig. 2 is a schematic diagram of the LED driving circuit of the present invention.

Fig. 3 is a schematic diagram of the MLP neuron network of the present invention.

Fig. 4 is a schematic diagram of the utility model using BP training algorithm.

Fig. 5 is a working schematic diagram of the neuron merging module of the present invention.

detailed description

The utility model will be described in further detail below in conjunction with the embodiments and accompanying drawings, but the implementation of the utility model is not limited thereto.

Example

As shown in Figure 1, a visible light communication system based on neuron network. Including: transmitting subsystem, transmitting subsystem, receiving subsystem. In the transmitting subsystem, the camera receives a real-time video signal and transmits it to the first liquid crystal display through the first programmable gate array; the first liquid crystal display shows the original video signal; After the video signal is modulated and processed with corresponding video signal processing technology, it is transmitted to the LED drive circuit; the LED drive circuit drives the LED lamp to emit visible light signals. As shown in Figure 2, it is a schematic diagram of the LED drive circuit. Through the T-shaped combination of the capacitance and the inductance of the BiasTee structure, the coupling of the DC signal and the AC signal is realized to ensure that the signal will not be lost in the LED.

Further, the optical signal reaches the receiving subsystem through the transmission subsystem; the transmission subsystem is 4 different channels in this embodiment. The optical signal is converted into an electrical signal by 4 photodetectors respectively, in the form of 4 data streams; the 4 data streams are filtered and amplified by the amplifier circuit and the filter circuit, and then the electrical signal is processed by the ADC analog-to-digital converter Quantize and convert to digital signal. Then through the neuron merging module, the 4-way data streams are merged to obtain the final data stream.

Further, the neuron merging module uses a three-layer MLP neuron network, and uses BP algorithm to learn and train the neuron network. As shown in Figure 3, it is a schematic structural diagram of a three-layer MLP neuron network; as shown in Figure 4, it is a BP training algorithm. The output state Y _i of the i-th neuron is:

${\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>$

In the formula, is the weighted value of the connection between the i-th neuron and the j-th neuron; V _j is the output of the j-th neuron; Q _i is the threshold of the i-th neuron; it is the activation function.

During training, a set of input values and corresponding expected values are given to the neuron network, and the connection weights are adjusted according to this set of training data. An error signal can be obtained by comparing the expected value with the output value of the forward propagation.

E(x)=[y(x)-F(x)] ² ,

Among them, x and y(x) respectively represent the input signal and the corresponding actual output signal; F(x) is the expected output signal. The gradient descent method is used for error backpropagation and weight correction, and the error signal reaches or falls below a set value through repeated learning.

${\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}\frac{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{<span\; class="notranslate">\; \&part;</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; ;</span>$

Among them, _ωij represents the connection weight value, and γ represents the learning rate. Through the training of the above equation, the connection weights can be adjusted to minimize the mean square error between the actual output of the neural network and the expected output. After BP algorithm training, for any input value, the neuron equalizer can give a relatively more appropriate output. Thus, equalization processing of the channel is realized. As shown in Figure 5, the trained neuron network combines n-way data into one data stream, and sets a threshold of 0.5 at the output of the neuron to generate binary codes. Wherein, in this example, n=4.

The final data stream is then converted into a video signal by the second programmable gate array, and then transmitted to the second liquid crystal display. By comparing the signals in the first liquid crystal display and the second liquid crystal display, the bit error rate of the system can be tested to verify the operation of the system.

The above-mentioned embodiment is only one embodiment of the present utility model, but the embodiment of the present utility model is not limited by the above-mentioned embodiment, and any other changes, modifications and substitutions made without departing from the spirit and principle of the present utility model , combination, and simplification should all be equivalent replacement methods, and are all included within the protection scope of the present utility model.

Claims (3)

1. the visible light communication receiving system based on neuroid, comprise: launch subsystem, transmission subsystem and receiving subsystem, it is characterized in that, described transmitting subsystem has: the first programmable gate array, LED drive circuit and LED lamp, and described receiving subsystem has: n photoelectricity testing part, ADC analog to digital converter, neuron merge module and the second programmable gate array; Described first programmable gate array, LED drive circuit, LED lamp, photoelectricity testing part, ADC analog to digital converter, neuron merge module and are connected successively with the second programmable gate array.

2. the visible light communication receiving system based on neuroid according to claim 1, it is characterized in that, described LED drive circuit comprises: information source, variable resistor, high-speed buffer, BiasTee module, DC current source and current-limiting resistance, and described information source, variable resistor, high-speed buffer, BiasTee module are connected successively with current-limiting resistance; The positive pole of described DC current source is connected with information source, the negative pole of described DC current source and BiasTee model calling; Described BiasTee module comprises electric capacity and inductance; One end of described inductance is connected with the negative pole of DC current source, and the other end of described inductance is connected with the negative pole of electric capacity, and the positive pole of described electric capacity is connected with high-speed buffer.

3. the visible light communication receiving system based on neuroid according to claim 1, it is characterized in that, described transmitting subsystem also comprises: the first liquid crystal display and video camera, and described receiving subsystem also comprises amplifying circuit, filter circuit and the second liquid crystal display; The first described liquid crystal display is all connected with the first programmable gate array with video camera, and photoelectricity testing part is connected with ADC digital to analog converter with rearmounted amplifying circuit by pre-amplification circuit, and the second programmable gate array is connected with the second liquid crystal display.