Modified OOK In DWDM-FSO Systems Under Atmospheric Turbulence Channel And Interchannel Crosstalk


 This paper presents designing and analysis using a dense wavelength division multiplexing free-space optical (DWDM-FSO) communication systems and shows the noise effects, interchannel crosstalk, and atmospheric turbulence in the weak and strong turbulence with an on-off keying (OOK) modulation. Numerical results show error floor occurs in the DWDM-FSO link using an OOK and adaptive detection threshold.


INTRODUCTION
Although linking communication in a dense wavelength division multiplexing free-space optical (DWDM-FSO) communication systems instead of the radio one allows ignoring license requirements, improving security, and transmitting data with broad-spectrum, it suffers from fluctuations in temperature and pressure that occur in the atmosphere regularly [1], [2]. In the case of such systems that detect directly with constant thresholds and on-off keying as well as modulate irradiance, it leads to appearing error floors at high-signal to-noise ratios (SNRs) [3], [4]. There are claims that one can avoid them through adapting the thresholds and on-off keying modulation (OOK) [5]. In the case of the limit when there is no noise when the decision threshold is continuously adapted to the changing received power (signal + crosstalk) it will be set at responsivity (signal average power + crosstalk average power) [where the average is over data but is changing depending on turbulence in either the signal case or the crosstalk case]. For that fraction of time that the crosstalk average power is bigger than the signal average power (regardless of whether it is the signal or the crosstalk's turbulence that causes this) then the data on the crosstalk wavelength is always recovered in preference to the signal data. For scintillation states within this fraction, the error probability bit-error-rate (-BER‖) must then be 0.5 (as signal and crosstalk data uncorrelated so crosstalk -guesses‖ the signal half the time). When the more usual situation that the signal average power is bigger than the crosstalk average power (regardless as to which case i.e. whether it is the signal or the crosstalk's turbulence that causes this) then the data on the signal wavelength is always recovered in preference to the crosstalk data. For individual scintillation states within this fraction of time the error probability (-BER‖) is then each 0. So we end up with an error floor for the average BER of 0.5 F+ 0 (1 F) = F/2 where that average is over all states of scintillation for the pdf arising from one Rytov variance and F is the fraction of time during which the crosstalk is greater than the signal. If we introduce a small amount of noise into the system, the thermal noise signal independent for simplicity that we will get deviations from this floor for lower signal powers but sufficiently high signal powers this small amount of noise can be neglected and the error floor retained. However, there is a flaw in this argumentation. Particularly, the definition of F follows that the fact that BER equals F/2 does not necessarily imply that BER does not depend on the received signal average power. Moreover, it is impossible to replicate the graphs with the floors they present. Specifically, in [6], the first error floor appears in Figure 2a. This illustration contains among others the dependences of BER on average received optical power in the cases that are the following: 1) there is neither interferer nor turbulence (Si); 2) there is turbulence in the signal but there is no interferer (TurbuSi); 3) there is the interferer but the turbulence is present only in the signal (TurbuSi, XT). This paper has been organized as follows. Section 2: presented system design of DWDM-FSO of 8 channels with two parts transmitter and receiver. Section 3: discusses the error floor in the case of adaptive threshold numerically. Section 4: we shed light on subtleties of an adaptive detection threshold calculation. Section 5: introduces the modified OOK modulation. Finally, section 6: show the simulation results of the quality factor and eye diagram for DWDM-FSO with nonreturn-to-zero (NRZ) and return-to-zero (RZ).

SYSTEM DESIGN
DWDM-FSO link can increase the number of channels and thus helps to enhance the technical capacity and use for long-distance data transmission. There are three types of WDM (wavelength division multiplexing) (WDM) that are commonly used: Coarse WDM (CWDM), Dense WDM (DWDM), and Broadband WDM. DWDM is a technology that multiplexes multiple optical carrier signals on a single medium by using different signals with huge channel capacity and the wavelength range from 1539 nm to 1565 nm is the most commonly located in C-Band and contribute to reducing bandwidth usage and can support capacity to reach Terabits per second and are easy to increase a data rate into the FSO [7]. Passive optical networks (PON) (i.e. the last mile connection between individual homes and companies) and the general network and gradually replace the copper-based on of network access technologies [8,9].WDM is the next generation of dissemination of FSO based access network which offers higher bandwidth [10,[11][12][13][14] With (WDM-PON), is set fixed wavelengths for each optical network unit (ONU), and thus more fully exploit the high transfer bandwidth available in the optical domain [12,13]. (WDM-PON) systems offer greater security; increase bandwidth and less loss compared to time division multiple accesses (TDM/TDMA) [6][7][14][15][16][17][18][19][20]. Figure. 1 Schematic representation of a DWDM using FSO link.

TRANSMITTER MODULE
In the transmitter section, we have used externally modulated transmitter in order to achieve stability. This also helps to reduce the chirps and non-linear effects [6,15]. Here, pseudo random bit sequence (PRBS) is used to generate digital data and NRZ pulse generator is used to convert digital signal into electrical signal. After that modulator mixes the electrical signal with the light source input signal and generate optical output signal which is then sent into the multiplexer. In our design, equally spaced of 100 GHz frequency separation of eight transmitters is used starting from 1550 nm and feeder fiber length 20 km [6,[18][19][20].

RECEIVER MODULE
In receiver section, PIN photo-detector is connected to the output to detect the optical signal and convert it to electrical signal and send it to low pass Bessel filter which pass the low frequency signal and discard high frequency carrier signal [2,6].

THE ERROR FLOOR IN THE CASE OF ADAPTIVE THRESHOLD NUMERICALLY
In what follows, we use the commonly used notations and present all values in SI units if not explicitly written otherwise. In the first case, the dependence is the following [2,6,[18][19][20][21]: In the second one, it has the form Finally, in the last case the dependence is the following [6,[18][19][20][21]: Actually, the paper [6] is written so vaguely that it is difficult to figure out exact values of input parameters used. We believe that to replicate the graphs depicted on Figure 2a of the paper [6] under consideration that correspond to the above-mentioned cases, the following values of the input parameters written in SI units are appropriate:

ADAPTIVE DETECTION THRESHOLD CALCULATION
The major challenge consists in taking integrals in the right hand sides of the equations (2) and (4) properly. To find correct values for the integral limits in these equations, one should examine the dependences that are the following as written in [6,15,16,[18][19][20][21][22][23][24][25]: where X h is the attenuation due to atmospheric turbulence [6,[18][19][20] Since   , 9725 . 0 performing integrations in the equations (2) and (4) Figure 3 is the plot of the dependence of gamma-gamma probability density on X h . On this plot X h changes from 12 10  to 3. From Figure 3, it follows that on this interval there are no regions of high gradients of GG p . Therefore, one can use a uniform grid performing integrations. To obtain Figure 3 we cover the range of P from with the uniform grid that consists of 72001 nodes. To figure out whether it is enough, we perform the same calculations on the uniform grid that consists of 36001 nodes and calculate the percentage of the maximal deviation. Both in Case 2 and Case 3, the deviations are much lower than 1 %. Hence, the chosen grid is dense enough. Remarkably, conducting a numerical experiment we arrived at the dependencies of BER on average received optical power for three above mentioned cases that match the ones presented in [6] much better than Figure 3 does (see Figure 5). Attenuation due to Atmospheric Turbulence ( hx) 0.9

MODIFIED OOK
The description of the experiment is the following. One can take advantage of the uniform grid with 20001 nodes and choose for the lower limit in the integrals the value which is close to 0 whereas for the upper one the value which is much greater than the first number. For instance, one can try 12 10  for the lower limit and 3 10  for the upper one. However, the arbitrary choice of limits can lead to the normalization of the gamma-gamma probability density function that does not equal to 1 exactly. Therefore, one can introduce the normalization constant A in this function [6,20,21]: From Figure 2a of the reference [6], it follows that the values of   BER 10 log at the average received optical power equal to -30 dBm coincide in Cases 1 and 2. Hence, one can perform calculations assuming that the constant A presented in the equation (7) is equal to 1 and estimate its correct value as  10 where  is equal to in Case 1 minus such the value one obtains in Case 2 assuming that 13 2 10   n C . We plotted the curve presented on Figure 3 that corresponds to Case 2 and

SIMULATION RESULTS
This part shows the simulation result for analyzing DWDM-FSO over the noises and atmospheric turbulence accentuated interchannel crosstalk [17][18][19][20][21]. DWDM free space optics using NRZ line codes and RZ as shown in Figure 6. This system is designed using optisystem version 7. These channels are then multiplexed using ideal multiplexer.
Initial frequency is 1550 nm and frequency spacing is 100 GHz. Modulation type are NRZ and RZ. The attenuation of the laser power in depends on two main parameters: Attenuation and Geometrical loss.

CONCLUSION
This paper presented design a system model of DWDM-PON/FSO which can provide with higher bandwidth and growing demand of traffic and analyzed DWDM-FSO channel which accentuated interchannel crosstalk, interference and noises and proved that the results of numerical calculations for error floor that relies on OOK modulation and adaptive threshold false.