Simulation of the Effect of Noise on an Optical Fiber System

Introduction

Fiber-optic communication is used to transmit information between points using infrared radiation in an optical fiber medium. Such fibers allow the propagation of light over long distances, are also immune to electromagnetic interference and allow control over the direction of light rays by refracting them with TIR. Despite the clear advantages of fiber optic communications over, for example, classical metal cables, their use does not completely eliminate the occurrence of noise. Noise is to be understood as any unwanted interference that may occur during the operation of the optical fiber. With respect to this information transmission mechanism, there are three types of noise that can affect the efficiency of light propagation. First, there is thermal noise, which occurs in the conductor as a result of thermal excitation by the random motion of free electrons. Second is shot noise, which results from fluctuations in the number of detectable photons as a result of their discretization. Finally, there is RIN noise, which is a noise of optical intensity normalized relative to the mean value.

In the present laboratory work, the effect of noise is determined on the propagation of light in a fiber optic system modeled in VPItransmissionMaker. Noise effect patterns are determined for two data rates, namely 2.5 Gb/s and 10 Gb/s. Each of the speeds is considered sequentially; the maximum fiber lengths at which the BER reaches 10^-9 are determined for them. In addition, the work proposes an estimation of the SNR for the two speeds, followed by a comparison of their maximum performance.

Procedure

In the present work, VPItransmissionMaker software was used to simulate fiber optic systems with high accuracy and to measure the performance of the constructed model through risk assessment and component selection. In the virtual environment, a fiber optic direct detection VPI was built according to the results of previous activities, shown in Fig. 1.

The setup parameters for the simulation were defined as follows:

• Global:
  • Time window = 256/bit rate default
  • Sample mode BW= 100e9 Hz
  • Sample rate default = 80e9 Hz
  • Bit rate default = 2.5e9 bps
• MZM:
  • Extinction Ratio = 30 dB
• Laser:
  • Frequency = 193.1 x 1012 Hz
  • Average Power = 1 mW
  • Laser linewidth = 200 MHz
  • RIN = -150 dB/Hz
  • Noise Bandwidth = 0.7*bit rate default
• Transmitter Filter:
  • Rise time = 0.25/bit rate default
• Fiber NLS_PMD:
  • Length = 30e3 m (= 30 km)
  • Dispersion = 16×10^-6 /m² (= 16 ps/nm/km)
  • Attenuation = 0.25 dB/km
• PIN Photodiode:
  • Responsivity = 0.8 A/W
  • Dark Current = 0 nA
  • Thermal noise= 10×10^-12 A/sqrt(Hz) (default value).
  • Shot noise = on
• Transimpedance Amplifier:
  • Impedance = 50Ω
  • BW = 0.7 * bit rate default
  • Filter order = 3
• Receiver Filter:
  • Filter Type = Low pass
  • Transfer function = Bessel
  • BW = 0.7 * bit rate default
  • Filter order = 3
• Electrical Amplifier:
  • Gain=3
  • Current Noise Spectral Density = 10×10^-12 A/sqrt(Hz)
• BER_OOK_Stoch:
  • Estimation method = gauss
  • Detector type = PIN
  • Dark current = 0
  • Shot noise = No
  • Thermal noise = 0
• Optical Power Meter:
  • Measurement mode = TOTAL
  • Output units = dBm

Simulations

2.5 Gb/s

Using the Green Guy simulations for the input data (Q = 20.35, BER = 2.17×10^-92, Power = -10.68), an Eye Diagram was obtained, as shown in Figure 2. Figure 3 and Figure 4 show the dependencies of BER on Power and Q-factor: from the graphs shown, it can be seen that when the BER and Power values are combined, they give a Q-factor value of 20.35. Meanwhile, the good, relaxed shape of the graph indicates high signal integrity. Figure 5 shows the dependence of BER on the Q-factor when the thermal noise increases by ten values for each step (10 A/√Hz), and Figure 6 shows the dependence of BER on Power. In both cases, the BER increases, indicating an increase in bit errors with an increase in the effect of thermal noise. The Q-factor, on the contrary, decreases, and therefore an increase in the rate of energy loss and an increase in the rate of oscillation damping is observed. That is, thermal noise has a destructive effect on the BER and Q values for optical fiber.

To test the effect of shot noise, the thermal was set constant (10×10^-12), and the shot noise was increased sequentially in increments of ten units from 0.1×10^-3 to 10×10^-3 W. Figure 7 shows the dependence of the BER on the Q-factor: when the shot noise increases, the BER decreases rapidly, while the Q increases. The decrease in BER is also confirmed by Fig. 8, the dependence of BER on power, at which the power increases. Consequently, an increase in shot noise as an indicator of photon fluctuation activity leads to a decrease in the number of bit errors and a decrease in the rate of energy loss with slower decay of oscillations (FOSCO, n.d.). This may indicate that increasing the number of photon fluctuations leads to more efficient information transfer with fewer errors, but this is not an infinite process, and the BER comes to a plateau with time.

Finally, to measure the effect of RIN noise on optical fiber transmission characteristics, RIN values were manipulated from -110 to -160 dB/Hz. A decrease in RIN corresponded to a drop in the spectral power of the noise density compared to the full power. Figure 9 and Figure 10 show the dependencies of BER on Q and BER on power with increasing noise RIN, respectively. It is easy to see that, in this case, the number of bit errors decreased while the rate of energy loss and the rate of oscillation attenuation fell.

Based on the data obtained, one can conclude that a decrease in RIN, an increase in shot noise, and a decrease in thermal noise led to a reduction in BER. Simultaneously, an increase in RIN, an increase in shot noise, and a decrease in thermal noise led to a decrease in BER led to a decrease in energy loss (Q-factor). These conditions lead to increased information transmission efficiency over fiber.

10.0 Gb/s

All the same, procedures were performed at a 10 Gb/s energy transmission rate and a fiber length of 20 km. Figure 11 shows the Eye Diagram for this speed: in this case, it is obvious that the Eye Diagram shape is much more stressed. The heterogeneity of the shape indicates changes in signal amplitude and time; that is, higher data rates led to larger signal errors. “BER vs. Received Power” and “BER vs. Q-factor” diagrams are shown in Figure 12 and Figure 13.

Fig. 14 and Fig. 15 show plots of the BER versus Q-factor (Q) and Received Power. It is noteworthy that no change in BER or Q is observed when thermal noise increases. In fact, the BER values increased sharply with the primary increase in thermal noise, and Q, on the contrary, decreased sharply. From this, one can conclude that the number of bit errors reaches a maximum plateau when the thermal noise increases, as well as the rate of energy loss.

Fig. 16 and Fig. 17 show the same dependencies with increasing fractional noise. In this case, an increase in photon fluctuation activity led to a reduction in bit errors and a reduction in the rate of energy loss. For both characteristics, BER and Q reached their minimum and maximum plateaus, respectively. This again suggests that an increase in fractional noise has a favorable effect on the quality of the transmitted information.

Finally, Figure 18 and Figure 19 show the change in patterns for 10 Gb/s transmissions as the spectral noise density decreased. In this case, the BER decreased sharply, and the Q increased sharply. Hence, reducing RIN was a predictor for reducing the bit error rate and reducing the energy loss rate.

Maximum Fiber Lengths

When the thermal loss is limited to 10×10^-12 A/√Hz, at a given BER of 10^-9 at 2.5 Gb/s, the received optical power value can be calculated from Figure 8. Given the slope, the received optical power value at BER = 10^-9, is -2×10^-10. Since it was used a standard fiber with a decay factor of 0.2 dB/Km, the fiber length for this speed is 10×10^-9 km. For a transmission rate of 10 Gb/s, where thermal loss is limited (Figure 17), the fiber length was 10×10^-12 km. In other words, the maximum fiber length should have been higher at lower data rates.

Conclusion

This paper aimed to simulate a fiber-optic VRI according to the given characteristics and determine the effect of various sources of noise on the transmission performance at different speeds. It was shown that the effect patterns of the three different types of noise were identical for different speeds. At the same time, it was determined that the lower transmission rate required the highest maximum rate and vice versa. Based on the results, it can be concluded that the signal transmission efficiency for 2.5 Gb/s was higher because fewer errors were detected.

Reference

FOSCO. (n.d.). Shot noise and bit-error-rate (BER) for coherent demodulation and delay demodulation. Fiber Optics for Sale Co. Web.