Analysis of Visible Light Communication Link’s Performance in Fire Smoke Environment

. This paper mainly studies the communication performance of visible light communication link in ﬁre smoke environment. Compared with the previous studies, this study no longer only considers a single wavelength and considers the inﬂuence of smoke and background light noise. Firstly, the variationof backgroundlight noise caused by sunlight during a day was analyzed,then signal-to-noise ratio (SNR) and bit error rate (BER) of the visible light communication link in ﬁre smoke environment with diﬀerent visibility were calculated, and ﬁnally, thevariationofreceivedpowerwiththeangle β betweentheLEDoptical axisandthecommunication link(fromLED toreceiver) was analyzed. The results show that, during a day from 07:00 to 18:30, background light noise ﬁrst increases and then decreases, BER changes in the same trend and reaches the maximum at about 13:00, while SNR changes in the opposite trend and reaches the minimum at about 13:00. When visibility is 100m, the maximum diﬀerence of SNR between 7:00 and 13:00 is about 19dB, and the maximum diﬀerence of BER is about 5.3E-10. At 7:00, the maximum diﬀerence of SNR is about 13dB between the visibility of 1m and the visibility of 100m, and the maximum diﬀerence of BER is about 1.44 E − 11, and received power decreases with the increase of β . In addition, SNR increases with visibility of ﬁre smoke environment, and BER changes in the opposite trend. Lastly, in order to enhance the performance of visible light communication in ﬁre smoke environment, this paper gives the transmitting power scheme.


Introduction
Currently, the application of visible light communication (VLC) is becoming more and more widespread. e new visible light communication technology based on LED has many advantages: it can meet high-speed communication while realizing lighting, is harmless to human eye health and does not cause radiation damage to the human body, is not easily subject to electromagnetic interference, and has good confidentiality. In addition, VLC technology also does not require radio spectrum certification, which can effectively solve the problem of spectrum limitation of radio frequency (RF) technology [1] and can also be used for indoor target localization. erefore, this technology is playing an increasingly important role in vehicle networking, underground mines, and other fields.
Although visible light communication has many advantages, it is easily affected by the atmospheric environment [2], and indoor fire smoke and outdoor fire smoke are one of the most important factors. Due to the short light wavelength and serious light scattering, the visible light communication link is easily blocked. In addition, the outdoor rain, fog, snow, and background light noise are also important factors to interfere with the visible light communication link. In order to improve the performance of visible light communication in a special environment, it is necessary to study the transmission characteristic of the visible light communication link in this environment. In 2013, Zhu et al. [3] studied the effect of smoke on free-space optical (FSO) communication and proposed that the attenuation coefficient decreases exponentially as the volume of the shell increases. In 2016, Chen et al. [4] studied the

Analysis of Background Light Noise
e background light noise P bg cannot be ignored in visible light communication in fire smoke environment, and the blackbody radiation model can be used to describe spectral radiance W(λ, T B ) [8,9]: where hp is Planck's constant, c is the speed of light, λ is the wavelength, k is Boltzmann's constant, and T B is the average surface temperature of the sun, usually taken as 6000 K. e global solar irradiance E global has a wavelength range between 0.3 μm and 4.0 μm, and it serves as the input for global spectral irradiance measurements. e peak spectral irradiance S peak is given by [8] e global solar irradiance W approx can be derived as [8] e spectral range of the receiver E det is given by [9] E det (λ) � where λ 1 and λ 2 are the boundary values of the optical bandpass filter, and P bg can be obtained as follows [9]: where T 0 is the peak filter transmission coefficient, which can reach 1, A is the area at the receiving side, and n is the internal refractive index.
Based on the measured global solar irradiance, the trend of P bg during a day can be derived as shown in Figure 1.
As can be seen from Figure 1, before 13:00, with the sun rising, P bg increases from −3 dBm (7:00) to 16 dBm (13:00). After 13:00, with the sun falling, P bg decreases to 3 dBm at 18: 30, and from the above analysis, it is found that P bg changes obviously during a day. erefore, the influence of background light noise on the performance of visible light communication in fire smoke environment cannot be ignored, and the background light noise caused by sunlight is usually Gaussian white noise, which can be removed by using denoising.

SNR and BER of Visible Light Communication in Fire Smoke Environment
According to the Lambert-Beer law, when the transmission distance is D, the received power P r is determined by the attenuation coefficient. e visibility (V) is obtained from Koschmieder's law [10]: where T th is the visual threshold, which usually takes 2%. e relationship between the attenuation coefficient β λ and V can be approximated as where λ 0 is the maximum spectral value in the solar band. For the visible wavelength band, it is derived that e coefficient q depends on the experimental data such as the size and distribution of the particles [11]: en P r can be obtained as follows: where A R is the received range, θ is the beam divergence angle, D is the transmission distance, and P t is the emitted power of the LED. β λ is the attenuation coefficient(A R � 10 cm, θ � 60°, and D � 10 m). SNR can be used as a measure of visible light communication's performance, and SNR can be expressed as the relationship between received power and environmental noise, as follows [12]: where R is the photoresponsivity of the photodetector at the receiver, σ 2 shot is the scattering noise variance, and σ 2 thermal is the thermal noise variance.
where q is electric charge, B is noise bandwidth, and I 2 is noise bandwidth current [12].
where k is the Boltzmann constant, T k is the absolute temperature, η is the capacity per unit area, A is the area at the receiver side, Γ is the FET channel noise factor, G is the open-loop voltage gain, g m is the FET transconductance, and I 3 is the bandwidth coefficient of noise as a percentage of the full ascending cosine. Using OOK optical intensity modulation, BER can be obtained as follows: Select the LED with 45 mW emission power; when the background light noise is maximum, the following is P r in different visibility environment.
As can be seen from Figure 2, P r increases with an increase in visibility. P r increases from 0.0122 mW (V � 1 m) to 0.01302 mW (V � 12 m) and increases by 6.7% and finally converges to a constant value of about 0.01302 mW. e literature [13] gave the smoke attenuation coefficient at different wavelengths, and we took the wavelength of 670 nm for the calculation and compared it with the measured results in that literature. Figure 3 shows the comparison between the measured and theoretically calculated values of the attenuation coefficient.
As can be seen from Figure 3, the calculated value of attenuation coefficient is basically consistent with the measured value, and the relative error is about 0.07%. Here only the influence of smoke particles and background light noise factors is considered, while the complexity of the actual measurement environment is not only influenced by these two factors but also by the measured attenuation coefficient which is often larger than the theoretical attenuation coefficients. To avoid this problem, sunny weather should be chosen for the actual measurement. Figures 4 and 5 show the variation of SNR and BER during a day in fire smoke environment with different visibility.
By analyzing Figures 4 and 5, it can be seen that SNR first decreases and then increases, and BER first increases and then decreases during a day. Before 13:00, continuously SNR  decreases and BER increases, and finally, SNR reaches the minimum and BER reaches the maximum during a day. After 13:00, SNR increases and BER decreases. In addition, it can be concluded that SNR increases and BER decreases with the increase of visibility. e visible light communication system can be used to adjust P t to cope with the fire smoke environment. When V is 12 m and D is 10 m, P r is calculated to be about 0.013 mW. In order to make the receiving power reach 0.013 mW at any time, P t can be adjusted according to the real-time visibility.
As shown in Figure 6, when P r is always 0.013 mW, the power of the LED transmitter is adjusted with visibility.
From Figure 6, it can be seen that, in fire smoke environment, when V is less than 0.02 km, the less visibility, the more P t is required, when V is more than 0.02 km, P t tends to a constant value and is almost no longer affected by visibility. At this time, the transmitter can adjust the value of P t according to different visibility, so that P r is always at 0.013 mW.

Path Loss and Power Loss in Fire
Smoke Environment Figure 7 shows a simple visible light communication scene in fire smoke environment. e visible light communication link's performance in fire smoke environment is not only affected by the background light noise but also by the absorption of smoke particles. As can be seen from Figure 7, the distance between the receiving side and the LED is D, the radius of the detector at the receiving side is r, the angle between the communication link and the normal of the receiving side is α, the angle of the communication link to the axis of the LED light beam is β, Ω r is the solid angle of the detector's receiving field of view, and the receiving area is A. e relationship between them satisfies the following equation [12]: In the device manual of LED, the axial light intensity I 0 is usually given, and we can obtain the total luminous flux at the transmitter F s [12]: where g s (θ) is the spatial distribution function of the LED lamp and Ω max is the stereo angle of the LED beam, which is related to the maximum half-angle of the beam as follows: We can get the total luminous flux at the receiving end F r [14]: e LED can be viewed as a Lambertian source, and its spatial distribution function is g s (θ) � cos m (θ). e Lambert radiation coefficient m can be obtained by m � (−ln 2)/ln(cosϕ 1/2 ) where ϕ 1/2 is the radiation half-angle of the LED, and let θ max � π/2, the path loss L L can be derived by integrating.
In addition, the power loss L P can be expressed as

Complexity
where S s (λ) is the power spectral density of the light source and S r (λ) is the power spectral density of the receiving side. According to equation (19), it can be derived that It is known that when the visual acuity function is consistent, L p � L L .
LED parameter tables usually give F s or I 0 of LED, g s (θ), and the normalized power spectral density S s ′ (λ). When F S and I 0 are interchangeable, S S (λ) can be found from S s (λ) � CS s ′ (λ), where C is the conversion factor, which can be derived from equation (16) [14]: where λ sL and λ sH are two endpoint values of the wavelength range of the visible band, respectively, and S r (λ) can be obtained by the following formula: An optical filter with the spectral response of R f (λ) is added to the photodetector, and P r is expressed as follows: where λ rL and λ rH are two end values of the optical filter [14]. e normalized radiation power spectral density waveforms and visible light filter waveforms are shown in Figures 8 and 9.
As can be seen from Figure 8, there are two peaks in the normalized radiation power spectrum density of the LED, which are distributed at 450 nm and 560 nm. LED is a white LED with yellow phosphor on a blue substrate, blue LED itself does not have a very high luminous flux, but after adding the yellow phosphor, the excitation energy is 8 times higher than the original. So this LED is more suitable for visible light communication. As can be seen from Figure 9, the filter has a filter range of 380 nm-780 nm, which only allows all wavelengths of visible light to pass through, effectively filtering out the nonvisible band of the spectrum.
β is calculated to be constantly changing, when background light noise is maximum, and Figure 10 shows the relationship between P r and β in fire smoke environment.
As can be seen from Figure 10, P r is gradually decreasing with the increase in β. When the LED beam axis coincides

Conclusion
is paper investigates the factors that affect the performance of visible light communication in fire smoke environment. By analyzing the change rule of background light noise during a day, the changing relationship between SNR/ BER and visibility during a day and the influence of β was obtained. It is concluded that, with the enhancement of external light, SNR keeps decreasing, BER keeps increasing, and the communication performance decreases. In addition, the received power decreases with the increase in β.
rough the above study, it is concluded that the effects of fire smoke and background light noise on visible light communication performance cannot be ignored, and it is crucial to improve the performance of LED transmitting equipment and the performance of photoelectric detection equipment at the receiving side. For the transmitter side, the paper gives the corresponding transmitting power according to different visibility, and the adjustment scheme can be applied in the actual fire smoke environment. For the receiver side, in practical applications, the filter can reduce the effect of background light noise by increasing the modulation frequency. Meanwhile, β should be minimized. In the next research work, the above scheme will be applied in practice, and in the complex fire environment, the automatic power regulation device is used to verify the above scheme. Table 1 shows the parameters used in this paper.

Data Availability
e data used to support the findings of this study are obtained from previous studies [9,10,12].

Conflicts of Interest
e authors declare that they have no conflicts of interest regarding the publication of this article.