Optical communications – Multiplex – Wavelength division or frequency division
Reexamination Certificate
2000-06-30
2004-03-02
Negash, Kinfe-Michael (Department: 2633)
Optical communications
Multiplex
Wavelength division or frequency division
C398S034000, C398S038000, C398S092000, C385S140000
Reexamination Certificate
active
06701089
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to multi-span Wavelength Division Multiplexed (WDM) fiber optic communication systems and more specifically to a WDM system using Erbium Doped Fiber Amplifier (EDFA) optical amplifiers.
BACKGROUND OF THE INVENTION
Wavelength Division Multiplexed (WDM) optical fiber telecommunication systems can have extremely high overall data capacity since each channel is capable of carrying a high data rate signal. These high capacity signals can be carried cost-effectively over many hundreds of kilometers if Erbium Doped Fiber Amplifiers (EDFA) are used to boost the power of the optical signal periodically. Thee is a growing requirement to increase the capacity of the existing communication systems.
FIG. 1
illustrates a typical unidirectional fiber optic communication system. A transmitter terminal
111
includes a number (n) of transmitters
102
,
105
,
108
each of which transmits one channel at a certain power that is adjusted by their respective variable optical attenuators (VOA)
103
,
106
,
109
. A multiplexer
110
is used to wavelength division multiplex a plurality (n) of channels
101
,
104
,
107
. A plurality (m) of optical fiber spans
114
,
116
,
119
and in-line EDFAs
113
,
115
,
118
,
120
couple the transmitter terminal
111
at a first location to a receiver terminal
121
at a second location which is remote from the first location. The input to the receiver terminal is coupled to a dispersion compensation module (DCM)
122
. A demultiplexer
123
is connected to the output of DCM
122
and outputs from the demultiplexer
123
are coupled to n receivers
124
,
125
,
126
. There is also an operations, administration and maintenance (OAM) system
112
which is connected to the transmitter terminal
111
directly and to all other network elements indirectly via an optical service channel (OSC)
117
. The OAM
112
is comprised of a processing element, memory such as random access memory (RAM), flash memory and a permanent or removable storage device such as a hard disk drive, a floppy disk drive, or a compact disc drive (CD). The optical service channel
117
is shown separate from the optical fiber for schematic purposes only. In reality, it is carried by the optical fiber. This setup is a well understood unidirectional optical fiber communication system.
One major problem in such an implementation as disclosed in
FIG. 1
is the non-uniform wavelength dependent gain profile of the EDFAs
113
,
115
,
118
,
120
, and further within any other optical device that may be included between the multiplexer
110
and the demultiplexer
123
. These problems, inherent to the currently utilized EDFA optical fiber amplifiers result in each channel within a particular WDM system having a different optical gain and a different resulting Optical Signal to Noise Ratio (OSNR). Hence, some channels could have a relatively low OSNR and low received power which, in turn, could lead to an excessively high Bit Error Rate (BER).
A significant challenge in carrying such multi-channel signals over many spans of fiber separated by boosting EDFAs has to do with the fact that the wavelength spectrum of the gain of the EDFAs is not flat. In fact, as shown in
FIG. 2
, because of the physical properties of the Erbium ions that provide the gain, the shape of the gain spectrum
201
changes from strong gain (about 23.5 dB at 1530 nm) to weak gain (about 21.5 dB at 1560 nm). In a long multi-span cascade of fiber spans and EDFA line-amplifiers, the nominal gain of the EDFA is set equal to the span loss, so that a nominal channel does not rise or fall in power as it propagates downstream. This non-ideal gain spectrum means that in a long multi-span cascade of fiber spans and EDFA line-amplifiers, some channels will have more gain than the average and will grow in relative power as the multi-channel signal propagates down the link. However, some channels have less gain than average, and so the power of that channel will decrease as the multi-channel signal propagates down the link.
The amount of gain provided by an EDFA is controlled by the amount of pump laser power that is applied to the Erbium doped fiber, and typically covers a range of 15 dB to 35 dB. The amount of output power capability of the EDFA is also influenced by the amount of pump laser power. For any given amount of pump power, there is a certain limit to the total power over all of the channels, with 15 dBm as an example of a typical value. This is a natural physical limit at which the pump photon flux is just sufficient to replenish the depletion of the Erbium population inversion by the high signal output power. As well as this natural physical limit on the total power capability, there can also be an additional lower limit applied by design. For a given number of channels, it might be useful to limit the total power out of the EDFA and launched into the optical fiber in order to avoid certain nonlinearities in the fiber. This total power control (TPC) mode typically is implemented by tapping off a very small but controlled fraction of the light at the output of the EDFA and monitoring that with a photodetector.
Since all of the wavelength channels can carry revenue generating traffic, it is of interest to ensure that all of the channels meet a certain standard of performance. In a digital system, Bit Error Rate (BER) is typically used as a figure of merit, and 10
−12
is a common objective for BER. One of the main influences which will degrade the BER of multi-span EDFA links is the noise provoked by the Amplified Spontaneous Emission (ASE) which is generated inside the EDFAs. The amount of ASE relative to the signal power is typically quantified by the Optical Signal to Noise Ratio (OSNR), defined as:
OSNR=Signal Power/(ASE density*BW
OSNR
)
where BW
OSNR
is the spectral band over which the OSNR is defined (for example 0.1 nanometers)
To optimize the OSNR of any given channel in a multi-span link; the input powers to each EDFA should be kept as high as possible at all of the amplifiers. This influences the design of multi-channel links where some channels will be increasing in power going down span, and some channels will be decreasing in power. The simplest case to consider is one in which all of the channels are initially launched at the same power. In the case of a channel which has more than average EDFA gain, it increases in power after that initial launch point, up until the receiver. With such high powers going into the EDFAs, that channel will have a good OSNR and will then have a good BER, provided that fiber nonlinearities are not provoked. However, a channel which has less than average gain will drop in power at every span as it propagates down-link. This channel will have a poor OSNR and thereby will have a high BER, which may not meet an objective link 10
−12
.
FIG. 2
shows one way to deal with this channel gain disparity
201
is to use a so-called gain-flattening filter inside the EDFA so that the spectrum of the net gain (Erbium plus filter) is flat
203
at a particular gain value, called the Design Flat Gain (DFG).
FIG. 2
shows the gain spectra at 3 different gains. Trace
201
is with gain above DFG, 22 dB. Trace
202
is with gain below DFG, 18 dB. Trace
203
, is with gain exactly at DFG, 20 dB. It is recognized that this shape-cancelling process is not perfect and so there is generally a residual ripple at the DFG. Although the EDFA can be design-flat at a single gain value, the losses of the fiber spans can cover a wide range. A simple way to deal with this is to set the DFG at a large value by design, and then ask the end user to add carefully selected attenuation to each of their individual fiber spans to bring the total loss (span plus attenuator) up to the DFG. Although this can work, the extra loss added to every span will degrade the performance of the system.
A more sophisticated way to get around this problem resulting from gain differences between channels, but without adding loss, is to use an equalization technique. Equalization is do
Goodwin John C.
Lee Keith Y. K.
Negash Kinfe-Michael
Nortel Networks Limited
LandOfFree
Over-equalization for multi-span wavelength division... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Over-equalization for multi-span wavelength division..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Over-equalization for multi-span wavelength division... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3202021