FABRY-PEROT LASER IN THE OPTIWAVE SIMULATION ENVIRONMENT

The aim of this article is to point out the possibility of solving problems related to a concept of a flexible hybrid optical access network. The entire topology design was realized using the OPTIWAVE development environment, in which particular test measurements were carried out as well. Therefore, in the following chapters, we will subsequently focus on individual parts of the proposed topology and will give reasons for their functions whilst the last part of the article consists of values measured in the topology and their overall evaluation.


Introduction
The motivation of passive optical network (PON) technology is to provide a cost-effective access mechanism to end-users, which is interference-free and bandwidth-abundant.The general PON topology consists of single mode fibres linking the optical line terminal (OLT) and the optical network units (ONUs) with an optical distribution network (ODN).Based on this, TDM-PON technology is already mature [1].
Since the end users begin to use Internet services such as HDTV, VOIP, gaming or video conference, their demands for the bandwidth will exceed 100 Mb•s -1 by the end of 2012.In 2009 in Japan, the number of users connected via optical fibre exceeded 17 million.Copper lines and wireless access methods are not able to meet these requirements and that is why the attention is turning to the use of optical fibres [2], [3].
Passive optical networks got their name because of the fact that they don't use powered devices in the outside infrastructures, i.e. don't use so called active amplifiers or splitters and hubs, elements providing splitting of the light output power into branches.Passive splitters are used instead of active elements and allow dividing signal among many users.Typically, the optical signal is divided among 16, 32, 64 or 128 users within the maximum distance of 10-20 km.The passive network's big advantage over the active one is the fact it can be expanded relatively cheaply.Time-division multiplexing and wavelengthdivision multiplexing belong among the most popular multiplex methods.The time-division multiplex method is used especially for short distances within metropolitan networks etc. while the wavelength-division multiplex o n e i s u s e d f o r l o n g d i s t a n c e s , s u c h a s f o r e x a m p l e backbone networks.
Each of these multiplex methods has its pros and cons.Time-division multiplex is able to serve only a small number of participants -with low speed, however, for reasonable prices.On the other hand, wavelengthdivision multiplex can serve a high number of participants but its price is much higher.
In light of these different properties, there is an effort to create a hybrid network that would have positive characteristics of both of these multiplexes and at the same time would meet the characteristics of the passive optical network [4], [5].

The Stanford University ACCESS (SUCCESS) initiative within the Photonics & Networking Research
Laboratory encompasses multiple projects in access networks.One of these projects is the Hybrid TDM/WDM PON or SUCCESS-HPON [6], [7].

Hybrid PON
Time division multiplexing (TDM) and wavelengthdivision multiplexing (WDM) technologies have their pros and cons and therefore, in order to obtain the best topology properties for signal transmission, combinations of these two multiplex methods are being developed.The disadvantages of TDM are the advantages of WDM.The time-division multiplex is not able to fully exploit the optical fibre possibilities, and the wavelength-division multiplex is quite expensive as for the implementation.Their unification brings about good transmission properties for favorable economic aspect of the project.As for remote operation between OLT and ONT, we use the WDM multiplex.It is then divided by the arrayed waveguide gratings (AWG) demultiplexer into individual wavelengths, which are then sent to the topology.A part of the network, which is behind the WDM demultiplexer, already uses the TDM multiplex.That is why the TDM demultiplexer is located as close as possible to the target area, which is served by the access network [8], [9].

A Topology Design in the Environment Optiwave
As for the transfer, our topology uses the L band and C band.The L band is used for downstream and the C band for upstream.Thanks to using dense wavelength-division multiplexing (DWDM) technology, we used 32 wavelengths with 100 GHz spacing in each band.The each channel's speed is 10 Gbit•s -1 , the topology total transfer rate is thus 320 Gbit•s -1 bidirectionally on a single fibre (full duplex).Every wavelength is then multiplexed using TDM for two ONUs.
The entire concept can be divided into several sections: transmitters, receivers, fabry perot (FP) laser feed, outside plant and elements before and behind the plant.In the following chapters, we will have a look at the setting and implementation of the individual parts [10], [11], [12], [13].

Transmitters
In the transmitting section, we chose the continuous wave laser with the output power 5,5 dBm.We used CW_laser_Array instead of several continuous wave lasers and it helped us to group all the 32 lasers together.The adjustment was relatively simple.Parameters were set to these values: initial frequency -186 THz, spacing -100 GHz, output power -5,5 dBm, number of outputs -32, noise -100 dBm.For the sake of clarity, the output from CW_Laser_Array was then multiplexed into a single beam using the AWG with the inner attenuation of 3 dB.This option can be or doesn't have to be used in a real connection.It would depend on the level of integration of each component.The signal located at the multiplex output is shown in Fig. 2. Continuous wave laser is not an ideal source of radiance for the FP cavity.According to the Optiwave documentation, the FP cavity is used in conjunction with a white light source that passes through the Gaussian filter and then feeds the FP cavity.In our case, it would not be possible because a white light source can't be used in connection with the TDM multiplex.We also tried to use other lasers such as Pump Laser, however, this one didn't work with the TDM multiplex either.The output light beam then goes to another subschema where it is divided into individual wavelengths and assigned through FP filters to FP cavities.As for the FP filters, we set the desired mean value of frequency (according to the wavelength we wanted to establish in the FP) and bandwidth which is sufficient for the proper function of amplitude modulation and which later doesn't affect the surroundings wavelengths.Experimental measurements enabled us to find out that an ideal method for feeding FP laser is to use the white light source in combination with the Gaussian filter.However, despite the fact that the signal appearing at the output laser was sufficient according to an optical spectrum analyzer (OSA), it was not sufficient for the proper function of the subsequent application of the time-division multiplex.
After the advent of the filtered beam, the FP laser binds to its wavelength, which is then amplified and radiated away.As for the simulation, this wavelength is at its output.Data signal is created using PRBS generator (Pseudorandom binary sequence generator) and is brought to a pulse generator input with a return-to-zero (RZ).In the next step, the signal is processed in a subschema called TDM transmitter.Figure 6 shows the inner design of a part of the implemented connection that operates the optical timedivision multiplexing (OTDM).After the signal is brought into the sub-schema, it is divided into N parts, where N is the number of channels created subsequently.After its division, the signal is brought into the Amplitude Modulation (AM) input.The signals are modulated using the same carrier frequency [13].
A g a i n , P R B S a n d R Z g e n e r a t o r s a r e c o n t r o l elements.It is also necessary to use a time delay before the modulated signals are merged into one using a power combiner.As far as the first channel is concerned, its delay is zero and the element is inserted for the sake of t ra ns p a re nc y .A s f o r t he s e c o nd c ha n ne l , i t s d e l a y i s defined according to the following formula: where B is a bit rate, A is an order of channels (counted from zero) and N is a number of channels [13].After TDM multiplexing is completed, the signal is ready for transmission.This sequential filtering, amplification and TDM multiplexing is done for all 32 channels.Using the AWG, all these channels are then merged into a single beam, which is connected with the multiplexer before the path.
Figure 7 shows the output signal from a part of the TDM transmitter.There is a power level of randomly generated pulses.Thanks to this random generating the graph shows 2 types of signal power level.From zero to 6,5 mW approximately and from 6,5 mW to 13 mW approximately.Each TDM transmitter channel has its o w n P R B S a n d t h u s t h e r e c a n b e a s i t u a t i o n w h e n generators generate a pulse at the same time.These pulses then have a higher power level than other pulses because their power is summed up.

Feeding FP Laser
FP lasers were used mainly because of their possibility to tune themselves to the incoming wavelength and because of their possible spectral width for WDM.Thanks to this feature, it is not necessary that continuous wave lasers are on the side of ONU .The continuous w ave lasers are located at the side of OLT.Thanks to this, the provider is allowed to manipulate the used wavelengths more easily.Therefore, we created a sub-schema called "cw_laser_array_upstream_feed_32_channels", which contains a continuous wave laser field, the AWG multiplexer and the EDFA amplifier [13].
We set the first frequency for the laser field to 191 THz and the subsequent spacing for the 32 channels to 100 GHz.The output power was set to 5,5 dBm.Outputs from the laser field were then multiplexed into a single beam and amplified.EDFA amplifier has a gain control setting.That is why we set the gain value to 5 dB.Of course, we also had to set the noise occurring in active elements.In our case, we used 3 dB.The central value of the noise frequency was set to 192,5 THz and the noise b a n d w i d t h t o 5 T H z .N o i s e w a s t h e n i n t h e w h o l e spectrum of 32 channels.From the economic point of view, it is much more expedient to multiplex the outputs from the continuous wave laser into a single beam first and then to amplify, than to amplify each channel separately first and then to multiplex.

Transmission Medium
As far as a simulation of the transmission medium is concerned, we chose on optical fibre with preset parameters from the Optiwave library.Since the aim of our concept was to transmit upstream and downstream over a single fibre, we had to choose a bidirectional version of this fibre.Length of the tested path was 2, 5 and 10 km at full-duplex speed 320 Gbit•s -1 on 32 channels, i.e. 10 Gbit•s -1 per channel.Basic parameters of the fibre corresponded to G.652.C/D fibre and we present them in the table below.In the simulation program, it is necessary for the right functionality of the bidirectional fibre to add optical delay into reverse direction.We put the optical delay before and behind the optical fibre, see Fig. 8.We set every delay to 1. Another important parameter was a number of iterations of the simulation program.This can be set in Layout Parameters/Signal Iterations X, where X is the total number of all optical delays in the path of a signal +1.In our case, the X was 3 (X=3).The optical delay is very important in a simulation because without it we would be able to calculate only one direction in bidirectional components.This is connected with the subsequent displaying on imaging elements where we must set the signal index.If we talk about the signal before the optical delay, the index is 0. If the signal i s a f t e r t h e d e l a y , t h e i n d e x e q u a l s t h e n u m b e r o f iterations -1.

Elements Before and Behind the Outside Plant
Since we used the FP lasers that need a feeding signal, the fibre has to transmit two beams of signals in a single direction.The first one is the downstream signal and the second one is the feeding for FP lasers on the side of ONU.That is why we merged these two signals into a single one using a multiplexer.We used a two-channel multiplexer where a value of the first channel was set to 187,5 THz and the second one to 192,5 THz.We needed the multiplexer to merge two beams comprising of several channels into two channels only and hence we had to adjust the width of the first as well as of the second channel of the multiplexer to 5 THz.In this case, an ideal filter for this was a rectangular one [13].
Fig. 11: Spectral characteristic on multiplexer output before line.

Receivers
The last important parts of the concept are the receivers.These are implemented in the sub-schema "Receivers".Using the AWG demultiplexer, a signal arriving at its input is divided into individual wavelengths.In order to maintain the right functionality of the NxN AWG, all the unconnected inputs must be connected to the optical zero.And to do this, there is an optical "nuller".The particular channels are then divided into two through the use of a power splitter.As for the first signal, no delay is applied t

o i t , w h i l e a s f o r t h e s e c o n d o n e , t h e r e i s a d e l a y a c c o r d i n g t o t h e f o r m u l a ( 1 ) . B o t h s i g n a l s a r e t h e n
modified in the same way.An avalanche photodiode (APD) is set to capture the signal with the highest level.
The APD dark current is 10 nA.The ASE noise is turned on as well in order to reach the most realistic connection.Low-pass filter is another important element.Its boundary frequency is set to 8 GHz.The filter removes the high frequency noise resulting from the signal passing through its path due to nonlinearities, dispersion etc.Since the high frequency noise does not meet the Shannon-Kotelnik theorem requirements, there is an effect called aliasing.This effect is undesirable and that is why it is removed with the help of low pass filter which can also be called anti-aliasing filter.
The sampling theorem is given by the formula: where f max is the maximum frequency contained in the signal sampling, and f n is the Nyquist frequency, thus the highest frequency which can be sampled properly at the given sampling frequency.
Figure 13 and Fig. 14 show connections without a low pass Bessel filter and with it, respectively.A 3R regenerator restores the shape and the time base of the signal and amplifies it.The last element, a bit error rate (BER) analyzer, tells us the exact properties of the signal after its passing through the entire topology.And we will analyse these properties in the subsequent chapter.

The Topology Evaluation
In this chapter, we are dealing with the precisely measured properties of the topology.Parameters tested are the following: the total path attenuation, the drop of the output power between the transmitter and the receiver, and the quality of the signal whose basic properties are the Q factor, the bit error rate and the eye diagram.
The topology will be tested in downstream and upstream, with the path length 2, 5 and 10 km, and with the speed 10 Gbit•s -1 and 1 Gbit per channel.In the following chapters we will test the first channel of the downstream and the upstream.

Attenuation
The total path attenuation is the sum of all the attenuations on the path from the transmitting laser to the receiver.We set the attenuation to 3 dB for all the elements on the path, the attenuation on the path itself is 0,25 dB/km.As for the upstream, we have to allow for the fact that the signal will pass through the line 2 times owing to the location of the FP laser feeding on the side of OLT, therefore we inserted the EDFA amplifier with a 5 dB gain behind the feeding lasers.The results are shown in the following tables: Tab.2: Downstream attenuation.Considering the topology structure, the measured attenuation values are all right.The attenuation was gradually increased together with the length of the tested path.As for the upstream, we could detect an increased attenuation due to the effect of the Fabry-Perot lasers, which need a light source feeding.With such a large number of participants, a difference between AWG and an ordinary splitter is very visible.AWG has an advantage over splitter because the attenuation does not increase with a number of its outputs, as it is in the case of splitter.Thank to this feature, the designed topology is much more flexible for the option of a possible increase of participants.

Drop of the Output Power Level
The drop of the output power level gives us the output pow er loss level between tw o measured points in the topology.For our topology, we chose these locations.

For the downstream:
• point A is located behind the multiplexer for feeding lasers of the downstream, • point B is located behind the demultiplexer splitting the signal of the downstream and feeding the signal for the upstream.
For the upstream: • point C is located behind the EDFA amplifier of the feeding signal, • point D is located at the end of the transmission path for the upstream.
Tab.4: Drop of output power on line 2 km.Q factor or quality factor gives us the quality of signal with respect to distance of signal from the noise.It covers all the noises, dispersions and nonlinearities, which deteriorate the signal quality and thereby increase the bit error rate.It follows that the higher Q factor, the higher signal quality.Q factor is defined according to the following formula:

Bit rate [Gbit
where v 1 is a logic level "1", v 0 is a logic level "0", σ 1 is a standard deviation of a logic level "1" and σ 0 is a standard deviation of a logic level "0".
As for the first channel of our topology, we measured out the following Q factor:  The Q-factor value is closely dependent on the BER value.If the Q-factor value is 6, the BER value is 10 -9 .The values measured in our topology confirm this relation.When decreasing the Q-factor value, the bit error rate was increasing [16].

2) Bit Error Rate
Bit error rate is one of the main indicators of the quality of optical connection.It is under the influence of the same parameters as the Q factor.Bit error rate gives us the ratio between the number of mistakenly received bE bits and the total number of the received p bits in dependence on time.The relationship is given by the formula: where v is bit rate and t is time of measurement.
As for the first channel of our topology, we measured out the following BER: The BER parameter values measured in our topology were found in the range from 10 -7 to 10 -12 , as expected.Professional literature specifies this value as the maximum and minimum BER value respectively necessary for the correct transmission over the topology.The BER parameter value was changing in dependence on the bit rate and the length of the tested route [16].

3) Eye Diagram
The eye diagram shows the superposition of all mutually overlapping bits in the signal.The eye opening indicates the differentiability of the logic 1 from the logic 0. The more the eye is wide open, the greater the differentiability is, because it's better signal noise to ratio.Other eyereadable parameters are: jitter (delay fluctuation), intersymbol interference (ISI) and time length of rising or falling edge [14] [16].Parameters of the particular eye diagrams correspond with the measured values, which we have already mentioned in the previous parts of this chapter.It is clearly seen that the Q-factor is lower when the route length is 10 km than when it is 2 km.On the other hand, jitter is increased due to the effect of dispersion.Nevertheless, the measured parameters are still normal and a signal transmission would be realizable.

Comparison of APD and PIN Diode Properties
Another experiment was to compare the quality of signal receiving as for the APD and PIN diode.The experiment confirmed our theoretical assumptions, namely that, thanks to its greater sensitivity, the APD diode has better BER as well as Q factor values.Therefore, we used the APD diodes for our concept.The following figures show the individual outputs from the BER analyzer of the APD and PIN photodetector.The measuring was performed at a transmission speed 10 Gbit•s -1 on the 10 km long path.Setting the photodetectors was very similar.Responsivity of both diodes was set to 1 A•W -1 and the dark current value was 10 nA.However, as for the APD diode, there was a possibility to set up two more parameters -the gain to the value 3 and the ionization ration to the value 0,9 [15].-86,879

Conclusion
Hybrid access networks are a very important solution that will be used profusely in the future, thanks to its key properties such as large data transfer rate for a large number of participants for favourably low prices.This article describes one of an immense number of ways how to design a network.We used tree architecture and modern components in dependence on wavelength such as FP lasers and AWG routers.All the simulations had the expected results.The bit rate of 320 Gbit•s -1 for 32 channels duplex is sufficient for the 10 km distance.At this transmission rate, the bit error rate ranged from 10 -7 to 10 -12 , which are satisfactory results.It follows that a hybrid WDM/TDM PON is a quality future proof solution offering a great potential.In the next article, we will focus on different types of coding, light sources, modulations, on their influence on transmission quality for the given topology and on a test concerning the maximum possible number of participants.

Fig. 9 :
Fig. 9: The spectral characteristics at the output of EDFA amplifiers.

Tab. 10 : 11 Tab. 11 :Tab. 12 :
Bit error rate on line 2 km.Bit error rate on line 5 km.Bit error rate on line 10 km.

Fig. 17 :
Fig. 17: Eye diagram showing the first channel of the topology with 10 Gbit•s -1 on fiber length 10 km.

Fig. 18 :
Fig. 18: Eye diagram showing the first channel of the topology with 10 Gbit•s -1 on fiber length 5 km.

Fig. 19 :
Fig. 19: Eye diagram showing the first channel of the topology with 10 Gbit•s -1 on fiber length 2 km.
Features of G.652 C/D fibre. Tab.1: Drop of output power on line 5 km.Drop of output power on line 10 km.