RS encoding method has excellent characteristics, strong correction ability and higher efficiency in the linear block code, the group code length is short with good performance [4]. RS (15, 9) encoding can be used to deal with burst errors in visible light communication. Based on the finite field GF (24), the block length n is 15 (2⁴ − 1), the message length k is 9, the parity length e is n-k therefore is 6, the maximum error correction capability t is 3, the minimum code distance d is 7.

The generate polynomial expression is:

The native original polynomial expression is:

Therefore, the resulting matrix is:

According to the above formula, the formula after encoding is:

where x is the original element, and t is the error tolerance, x^ru(x)mod g(x) represents the remainder of x^ru(x) divided by g(x).

The native original polynomial expression can be calculated by the regenpoly instructions in Matlab. In the FPGA encoding circuit is designed based on g(x) multiplication or division method. The structure is realized by the current shift register structure, in which the adder is exclusive OR and not-carry, multiplier into the form of modulo two sums. The algorithm is constructed by the function [5]. The coding structure is shown in Figure 2.

Designed with Bias-T driver circuit, with ultra-high bandwidth, and can achieve analog digital dual-signal modulation. Light source drive circuit shown in Figure 3, the FB pin of constant current bias chip is the voltage loop feedback pin, combine with its peripheral circuit for constant voltage regulation, protecting LED and so on [6] [7]. ISP pin is the current feedback resistor of the positive connection terminal, can protect short circuit. ISN is the current feedback resistor negative connection terminal, the Bias-T drive circuit make light source work in the linear work area, the electrical signal superimposed on the DC bias.

According to the AC frequency of capacitor and inductance, fC = 1/(2πRC)

and fL = R/(2πL), the light source and load in series. Capacitance, assembling capacitor and inductance and series 50 Ω matched microstrip lines, to reduce the return loss to expand the bandwidth to ensure the signal fidelity.

The characteristics of the amplifier circuit shows that the first level of amplification is particularly important, while the noise is also amplified, so the circuit requires a low noise amplifier chip. The chip selection should take into account the characteristics of high bandwidth and in-band gain stability. So the circuit uses a high gain bandwidth, low noise voltage feedback amplifier (OPA2846), receiver amplifier circuit shown in Figure 5. The amplified signal still has the form of an analog signal, so the signal is sampled by the modulo sampling circuit module to provide a larger signal dynamic range for the signal decision. The AD9280 chip has the characteristics of high speed processing ability and low power consumption. The sampling circuit is composed of AD9280 and the attenuation circuit.

The signal received by the experiment still has the form of an analog signal, so the data is processed according to the sampling data provided in the previous chapter. In order to provide a larger dynamic range for the signal, and to ensure the correctness of the subsequent signal processing, the sum of the adjacent sampling pulses is summed in the FPGA to prevent two or more miscarriage of justice in one frame of data. Improve the reliability of communication systems. The implementation process is shown in Figure 6.

The design of the RS decoder consists of a syndrome calculating circuit, a key equation solving circuit, a Chien search circuit and a Forney algorithm circuit and so on [8], the logic circuit is shown in Figure 7. First, for the solution of syndrome calculating circuit, and the value of the syndrome S is actually calculated by the receiver symbol polynomial R(x), each value of the generator polynomial g(x) is substituted into R(x) to find the polynomial. The circuit is parallel to the 6-way solution to syndrome. On the error value detection, if the value of syndrome calculating circui is 0, there is no error, otherwise there is error. The finite

field multiplier in Figure 7 uses MATLAB to calculate the constant coefficient multiplier, which is efficient and can improve accuracy, and the result can be saved as a.v format.

If the symbol is wrong, then the error from this step to find. For the solution of key equations, the key equation is S(x)σ(x) = ω(x)modx^2t, σ(x) is the error position polynomial, and ω(x) is the error value polynomial. In the operation, the data is initialized and the current syndrome is updated by the internal memory of the calculation unit [9]. One iteration for each clock cycle, and the result can be outputted until the iteration ends. Inversion of the results yields the coefficients of ω(x) and σ(x).

After the above calculation, the error position and the error value are required. The error position is actually the value of σ(x). As shown in Figure 8, the Chien search method verifies whether σ(x^l) (l = 0, 1, ・・・, n) is 0, that is, find the value by verifying each r_n−1 to determine the location.

The final output is the correct source code, the input data and error data through the XOR logic operation, the correct source data can be got. Debugging delay circuit and XOR operation circuit, it is necessary to delay the error data, only the correct delay period makes the data not aligned, the final decoding data can be obtained.

Based on FPGA of EP4C515F17C8N to achieve the pulse of the hardware mod-

ulation, the original signal is first sent to the register cache, driven by the drive circuit to the light source to send out; receiver sampling circuit using AD9280 chip of AD, the maximum sampling frequency of 32 Msps, using logic analyzer to simulate to meet the signal dynamic range of the pulse decision algorithm. The test results are shown in Figure 9.

When the driving circuit module is tested, the square wave signal is generated by a signal generator (SUIN TFG6060). The voltage is 5 Vpp, the signal waveforms are kept good at 20 KHz to 12 MHz, but the square wave signal from 6 MHz starts to narrow at the bottom and the top fluctuation starts to slowly increase. The top and bottom distortion is more serious from 12 MHz, so the test shows that the drive circuit can transmit the OOK signal within 12 Mbit/s and achieve the desired requirements in this article. The oscilloscope (RIGOL DS2202) test results are shown in Figure 10.

The amplified circuit is the secondary amplifier circuit at the receiving end. Since the signal measured by the photodetector still has the form of the analog signal, and the signal is almost unchanged through the transimpedance module, the function signal generator is used to generate a sine wave signal for testing; the signal have a gain effect from 6 KHz gradually when changing the frequency during the test, but the waveform is poor; the gain is reduced when the frequency reaches 50 MHz, the effect relatively is best at 35 KHz - 22 MHz. As shown in Figure 11, the signal gain waveforms with 5 v and 20 MHz respectively is given, and the gain with 4 V respectively at 20 MHz. From the test results, the linear stability and gain effect are good.

Test the performance of the system by testing the eye. In this test method, use the FPGA to generate a way of 300 Kbit/s pseudo-random signal, the modulation method use the key switch modulation. The pseudo-random signal generated by FPGA is loaded into the LED drive circuit, and the oscilloscope is connected directly to the end of the waveform test point and the clock output point. The measured eye diagram of system as shown in Figure 12, it can be seen that the eye pattern of receiving end is significantly more open and no overlap between signals, which fully illustrates that the system has good communication ability.

Finally, the BER performance of the system is analyzed and the measurement is carried out by using the illuminance meter and the AV5233C error detector. In the test, the angle of the lens is adjusted to match the light intensity of the table, then the intensity of the optical signal is moderate by adjusting the position of the PIN, and the receiving terminal can synchronize signals. The illumination and BER of VLC system is shown in Table 1.

From the measurement, when the distance is less than 0.5 meters, the illumination is 42 KLux, photodetector to saturation; at a distance of about 15 meters, the illumination reached 55 Lux, began to appear error, reaching 10 - 8 W. When the distance is more than 25 meters, the error rate is the number of 10 - 6, so we can see that the reliability of the system is good.

Test chart of video image transmission system is shown in Figure 13. 1) shows the test results with background light close-range, using small condensing lens whose focal length is 110 mm and the diameter is 40 mm, it can be seen from the chart display, can display the clear video image in real-time; 2) shows the test results without background illumination, it can be seen that the white LED can provide good lighting for monitoring, but also can realize the video image monitored transmission; 3) shows remote testing, using the lens whose diameter is 102 mm and focal length of 240 mm, the farthest distance can reach 4 meter, and the video can also display at more than 4 m, but the display is not stable at the same time, the video display is not color but gray.