Note: Descriptions are shown in the official language in which they were submitted.
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
OPTICAL NULTICHANNEL SYSTEM
BACKGROUND
Applicant's invention relates to telecommunication
networks that use optical fibers.
Optical systems and circuits have become more and
more important for data communication systems. Optical
fiber networks are particularly useful in
telecommunications systems because optical fibers have
large transmission capacity without electromagnetic
interference and ground loop problems.
With the demand for transmission capacity
increasing, as for example in broad-band multimedia
telecommunications, there is a need for optical
multichannel systems. Optical multichannel systems will
probably change network design strategies during the
coming years. By using multichannel techniques,
increased transmission capacity and flexibility can be
realized on existing fiber cables without increasing
modulation speed or adding more complex control
functions.
A ring architecture is common for communication
networks, and when using optical fiber cables, the ring
usually has one fiber carrying normal traffic in one
direction and another fiber carrying the same traffic in
the other direction for protection against traffic loss
due to a fiber break. In this way, each node in the
network can be reached in two separate ways with just
one optical fiber cable, so if a fiber break occurs in
one direction, the traffic can be transmitted across the
other fiber in the other direction.
Optical fiber amplifiers are usually provided to
compensate for signal attenuation. Erbium-doped fiber
amplifiers (EDFAs) are the most common and so far the
best, but there are also other candidates. While such
amplifiers compensate for signal attenuation, they can
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
2
also magnify their own spontaneous emissions, leading to
problems in ring architectures. In particular,
amplified spontaneous emission (ASE) from the optical
amplifiers can circulate in the loop if no special
measures are taken to prevent this. The ASE circulation
leads to saturation, higher noise level and
oscillations. This problem appears not to be easily
solved with optical filters.
One way of providing the desirable signal
protection is described in A.F. Elrefaie et al., "Fiber-
amplifier Cascades in 4-fiber Multiwavelength
Interoffice Ring Networks", Proc. IEEE LEOS 1994,
Optical Networks and Their Enabling Technologies,
pp. 31-32 (1994). One limitation of this system is the
circulating ASE problem.
Circulating ASE and techniques for coping with it
are described in K. Bala et al., "Cycles in Wavelength
Routed Optical Networks", id., at pp. 7-8. Although
this system solves the problem of ASE by breaking
oscillating cycles, it does not provide the desirable
protection of a bi-directional bus network.
A scalable, multi-wavelength, multi-hop optical
network and various aspects of optical multichannel
systems are described in C.A. Brackett et al., "A
Scalable Multiwavelength Multihop Optical Network: A
Proposal for Research on All-Optical Networks", IEEE J.
Liahtwave Technol. vol. LT-11, pp. 736-753 (May/June
1993). This system focuses on scaling the number of
channel wavelengths and reconfiguration of wavelengths
and does not solve the problems of circulating ASE.
Other aspects of optical multichannel systems and components useful for such
systems are described in C.M.
Miller et al., "Passive Tunable Fiber Fabry-Perot
Filters for Transparent Optical Networks", Proc. 3rd
IEEE International Workshop on Photonics Networks,
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
3
Components, and Apnlications, Atlanta, Georgia (1993);
V. Mizrahi et al., "Four channel fibre grating
demultiplexer", Electronics Letters vol. 30, no. 10
(1994); and J.E. Baran et al., "Multiwavelength
performance of an apodized acousto-optic switch", Proc.
OFC194 Paper No. TuM5. These publications describe in
detail various filtering arrangements that can be used
in the nodes of an optical network for wavelength
division multiplexing (WDM), i.e., to select channels of
different wavelengths.
U.S. Patent No. 4,973,953 to Shimokawa et al.
discloses a data transmission system which transmits
data between nodes linked in ring form. Each node
includes two transmitter circuits for simultaneously
transmitting data and frame signals in opposite
directions and a control circuit for detecting t7at?
transmission faults and supervising data transmission of
the system according to the location of the fault in the
system. Nevertheless, the Shimokawa system is not an
optical system but is all electrical, and the Shimokawa
system is not a multichannel system but is only a single
(electrical) channel system. Also, the Shimokawa system
terminates the electrical signal, processes it, and re-
transmits it, in each node. Thus, the Shimokawa system
does not provide direct node to node communication.
U.S. Patent No. 4,704,713 to Hailer et al.
discloses an optical fiber ring network which is capable
of operating if one of the nodes in the ring fails.
Each node is able to diagnose a failure of its own main
receiver or of a transmitter in the immediately adjacent
upstream node. In either case, the node switches from
its main receiver to its alternate receiver to bypass
the upstream node while enabling the remainder of the
ring to continue functioning. In the Hailer system, two
wavelengths are used for sending traffic one and two
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
4
nodes downstream. Thus, the Haller patent does not
describe an optical multichannel system as that term is =
commonly understood. Moreover, the Hailer patent is
directed to solving node failure problems, not cable
breaks.
European Patent Publication No. EP 0 487 392 to
Dequenne discloses a bi-directional optical multiplexing
system which uses different wavelengths for transmission
in two directions around an optical fiber loop. The
nodes communicate bi-directionally with four wavelengths
through the same fiber. Nevertheless, the channels are
terminated in each node.
SUMMARY
In accordance with Applicant's invention, there is
provided a communication network having a bi-directional
bus architecture, in which each node has at least one
on/off node switch, i.e., a switch that permits or
blocks transmission around what would otherwise be a
ring. At any given time, at least one node switch is
off, thereby avoiding problems with circulating
amplified spontaneous emissions. If a fiber break
occurs, the node switch (or node switches) in the node
adjacent to the break and on the same side of the node
as the break is (are) switched off, and the node switch
(or node switches) switched off before the break is
(are) switched on. This permits the network to operate
largely as before. Thus, Applicant's invention provides
a bi-directional bus in which a protection cable is
drawn the shortest way, leading to a ring structure.
In one aspect of Applicant's invention, an optical
fiber communication network for transmitting information on wavelength
channels comprises a cable having two
optical fibers; a plurality of nodes connected by the
cable; and a plurality of node switches, one for each
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
node. The nodes are arranged in a ring, and the nodes
and cable form a bi-directional bus. Every node switch
but one is in an on position permitting transmission of
information, and one node switch is in an off position
5 blocking transmission of information. The network
further includes a device for detecting a break in the
cable and a device for switching, in response to
detection of the break, the node switch that is in the
off position to an on position and the node switch in a
node adjacent to the break and on the same side of the
node as the break to an off position.
In another aspect of Applicant's invention, an
optical fiber communication network for transmitting
information on wavelength channels comprises a cable
having two optical fibers; a plurality of nodes
connected by the cable, the nodes being arranged in a
ring and the nodes and cable forming a bi-directional
bus; and a plurality of node switches, one on each side
of each node, wherein a first node has a first switch on
one side in an off position and a second node adjacent
to the first node on the same side as the first switch
has a second switch on that same side in an off
position, the first and second switches blocking
transmission, and every other switch in an on position
permitting transmission of information. The network
further comprises a device for detecting a break in the
cable and a device for switching, in response to
detection of the break, the first and second node
switches that are in the off position to an on position
and the node switches adjacent to the break on both
sides of the break to an off position.
In such networks, each node may include at least
two receivers and one transmitter for a respective
wavelength channel, and optical pre-amplifiers. Each
network may further include a wavelength reconfiguration
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
6
device for permitting re-use of a wavelength, the
wavelength reconfiguration device reducing a number of
wavelengths used in the network from M(M-1)/2 to
(M? 1)/4, where M is the number of nodes. The wavelength
reconfiguration device may include an add-drop filter
that is one of an optical circulator and a Fabry-Perot
etalon, an optical circulator and a fiber grating, two
three-port Fabry-Perot etalons, and an acousto-optic
transmission filter.
In another aspect, Applicant's invention provides a
method of transmitting information on wavelength
channels in an optical fiber communication network
having a number of nodes and a cable including two
optical fibers, the nodes being arranged in a ring, the
nodes and cable forming a bi-directional bus, and each
node including one node switch. The method comprises
the steps of setting every node switch but one in an on
position permitting transmission of information; setting
one node switch in an off position blocking information
transmission; detecting a break in the cable; and when a
break is detected, setting the node switch that is in
the off position to an on position, and setting the node
switch in a node adjacent to the break and on the same
side of the node as the break to an off position.
In another aspect, Applicant's invention provides a
method of transmitting information on wavelength
channels in an optical fiber communication network
having a number of nodes and a cable including two
optical fibers, the nodes being arranged in a ring, the
nodes and cable forming a bi-directional bus, and each
node including a node switch on each side of the node.
The method comprises the steps of setting a first switch
on one side of a first node in an off position and a
second switch in a second node adjacent to the first
node on the same side as the first switch in an off
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
7
position, the first and second switch blocking
transmission of information; setting every other node
switch in an on position permitting transmission of
information; detecting a break in the cable; and when a
break is detected, setting the first and second node
switches that are in the off position to an on position
and the node switches adjacent to the break on both
sides of the break to an off position.
Each method may further comprise the step of re-
configuring wavelengths in the network to permit re-use
of a wavelength to reduce a number of wavelengths in the
network from M(M-1)/2 to (M? 1)/4, where M is the number
of nodes. The re-configuring step may include the step
of add-drop filtering that is performed by one of an
optical circulator and a Fabry-Perot etalon, an optical
circulator and a fiber grating, two three-port Fabry-
Perot etalons, and an acousto-optic transmission filter.
BRIEF DESCRIPTION OF THE DRAWINGS
Applicant's invention is described below in more
detail with reference to preferred embodiments, given
only by way of example and illustrated in the
accompanying drawings, in which:
Figure la is a diagram of a network in accordance
with Applicant's invention;
Figure lb is a diagram of a network permitting
wavelength re-use in accordance with Applicant's
invention;
Figures 2a-2d depict other networks in accordance
with Applicant's invention;
Figure 3 illustrates a device for determining
whether or not a wavelength channel is present;
Figures 4a-4f illustrate examples of add-drop
filters for a network; and
Figures 5a-5b depict alternate arrangements for a
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
8
three-port Fabry-Perot etalon filter.
DETAILED DESCRIPTION
Applicant's network reflects a flexible bi-
directional network architecture (FBDNA) that solves the
problem of circulating ASE at the same time as it
maintains a ring structure. It also provides the
desired protection against transmission loss due to
cable breakage without wasting any fiber or number of
wavelengths compared with a uni-directional ring. (A
bi-directional bus requires fewer wavelengths than a
uni-directional ring). It also is an economical
solution since it requires relatively few and not-so-
advanced components compared to other solutions.
Figures la, lb illustrate network structures 10
that are built as loops of two optical fibers 5, 6, each
loop of optical fibers 5, 6 optically forming one bi-
directional bus for transmitting wavelength channels
X1 - X6 from node to node. Figures la, lb each show
four nodes 1, 2, 3, 4, although it will be appreciated
that more or fewer nodes could be used. Each of the
nodes 1, 2, 3, 4 contains an on/off switching device 20,
and all but one of these switches are set in a
transmitting (on) position. Figures la, lb show
expanded views of node 1, which includes one switch 20.
The bus starts and stops at the one blocking (off)
switch, and thus no problems with circulating ASE can
occur. The switching device 20 can be a conventional
optical switch or a pair of optical amplifiers that can
be turned on and off.
Each node also includes means for selecting a
desired wavelength channel. The selectors, or filters,
may be wavelength division multiplexers (WDMs), or
combinations of an ordinary optical fiber coupler and an
optical filter. (The latter precludes re-use of
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
9
wavelengths).
Each node also typically includes, for each
wavelength channel, a respective receiver R for each
fiber 5, 6. As indicated in Figures la, lb, node 1 has
six receivers R and six WDMs 22, 24, 26, 28, 30, 32,
three of each for each fiber 5, 6, and thus it can be
seen that node 1 can receive three of the six wavelength
channels X1 - X6. The desired signal would reach only
one of a node's two receivers at the same wavelength
since the signal would be present in only one of the two
fibers 5, 6.
Standard software or switches can be used to choose
the right receiver, the details of which are not
essential to the present invention and will not be
further described here.
Each node further includes one or more transmitters
T that are coupled to both optical fibers 5, 6 by either
WDMs or ordinary couplers, enabling the transmitters T
to send in both directions. As indicated in Figure la,
node 1 has three transmitters T that are coupled to the
fibers 5, 6, by two couplers 34, 36. Figure lb shows a
network that permits wavelength re-use with six
transmitters T coupled to the fibers 5, 6 by couplers
34, 36.
The transmitters T may be semiconductor diode
lasers or other suitable light sources. Light-emitting
semiconductor diodes (LEDs) are generally not suitable
transmitters since they have broad wavelength spectra
and are therefore difficult to separate from each other.
If LEDs are used, only a few channels can be transmitted
since they fill up the fiber's available spectrum, and
special types of filters must be used. Semiconductor
lasers that are suitable transmitters include
distributed feedback (DFB) lasers and distributed Bragg
reflector (DBR) lasers. For narrow channel spacing, the
CA 02218089 1997-10-10
WO 96/32787 PCTlSE96/00441
lasers must be nearly chirp free. A good way to prevent
chirp is to use integrated electro-optic modulators with =
the lasers.
The receivers and transmitters might be standard
5 components or more advanced devices such as coherent
transceivers that have been described in the literature
referred to above. One of the benefits of coherent
systems is the possibility of electrical filtering.
Figures 2a-2d show networks 10' in which the nodes,
10 for example node 1', include optical pre-amplifiers OA
for compensating signal attenuation in the optical
fibers. (A small wavelength separation between the
channels is desirable since the gains of the optical
pre-amplifiers typically vary with wavelength; this
ensures a nearly uniform gain for each channel.) In the
network of Figure 2a, for example, the dominant noise in
each receiver R is not thermal noise but noise due to
beating between the optical signal and the ASE. Under
such conditions, decreasing the received signal power is
not crucially important. Exit fibers leading from each
pair of wavelength-channel selectors (e.g., WDMs 22, 28)
are coupled into respective single receivers R. instead
of into two separate receivers, for each channel. This
embodiment requires no switch and/or software for
choosing the right receiver.
Each node further includes means for detecting a
cable break. A cable break between two nodes can be
detected as a loss of ASE coming into a node from a
direction or a loss of all wavelength channels from a
direction. These two main ways of detecting a cable
break are then used in controlling the network,
depending on whether node control is centralized or
local. Figures la, lb, 2a, 2b depict networks that are
centrally controlled. In centrally controlled networks,
signalling between nodes is controlled by a network
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
11
management circuit 40. Figures 2c and 2d depict locally
controlled networks. In locally controlled networks,
signalling between nodes is controlled by electronics
within each node.
For a network having centralized node control, it
is first detected which channels are lost, and then it
is determined where the break has occurred. Next, the
blocking (off) node switch is opened (turned on), and
the node switch that is on the same side as the cable
break is closed (turned off). As depicted in Figures
la, lb, 2a, and 2b centralized node control networks
must include one two-way switch in each node.
One way of detecting if a channel is present or not
is to superimpose a pilot tone on the channel. As
illustrated in Figure 3, a node can then easily extract
the pilot tone that identifies the channel by means of a
simple coupler and detector, including some simple
electronics. The detector in Figure 3 detects the
absence of optical power or the presence of the pilot
tones modulating the channels in one of the fibers 5, 6
based on only a small part, e.g., 5%, of the optical
power in the fiber. A technique for detecting a pilot
tone is disclosed in G. R. Hill et al., "A Transport
Network Layer Based on Optical Network Elements", IEEE
J. Lightwave Technol., vol. LT-11, pp. 667-676 (May/June
1993), which is incorporated here by reference.
Instead of a pilot tone, a separate wavelength
channel can be used for detecting a cable break. The
separate channel can be a low-bit-rate channel, so that
no optical filter is needed to detect the channel.
Another way of detecting a cable break is by
detecting the ASE or the optical power arriving at a
, node. If a node loses all channels, i.e., the power
coming into the side of the node at which the switch is
disposed is lost, the switch in that node turns off, and
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
12
the node starts sending a pilot tone in one or more of
the channels for telling the node that was originally
switched off to turn its switch on.
In locally controlled networks such as those =
illustrated in Figures 2c and 2d, decisions are made at
every node, and the status of the entire network does
not have to be known to take action. Figure 2c shows a
network having locally controlled nodes, with one switch
S in every node and a power or pilot-tone detector D
disposed in front of each switch. A detection result of
each detector D is sent to a respective switch S via an
electrical connection 7. Each detector D may take the
form illustrated in Figure 3.
The status of the network before a break occurs is
that one node has its switch S in a blocking (off)
position and all the other switches are in a
transmitting (on) position. If a node loses all
channels coming into the side of the node that the
switch is on, then the switch in that node shuts off.
The node that was originally shut off no longer detects
all channel signals from its on side but detects channel
signals coming into its off side from the node that has
shut off. Therefore, the node that was originally shut
off turns its switch on. The connections between the
nodes are in this way re-established.
For example, referring to Fig. 2c, assume that
switch S in node 1' is initially off to prevent ASE for
circulating. If a cable break occurs between nodes 3'
and 4', the detector D in node 4' will detect the break
and a signal will be sent along connection 7 to cause
the switch S in node 4' to turn off. Node 1' will
detect channels from nodes 2' and 3' from its on side
and detect a channel communicated from node 4' from its
off side, causing the switch S in node 1' to turn on.
Locally controlled networks such as shown in
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
13
Figures 2c, 2d require two-way switches S and a power,
or pilot-tone, detector D in front of the switch in
every node. Such networks will not protect against
breaks in only the fiber not having the detector D. To
protect against a single fiber break, i.e., a break in
just one direction along fiber 5 or fiber 6, a node can
be configured according to Figure 2d, which shows a
network having nodes that each include two switches S
and two power, or pilot-tone, detectors D, each detector
being connected in front of a respective switch.
Before a break occurs, a first node has the switch
on its right side shut off and the node immediately to
the right of the first node has the switch on its left
side shut off. All other switches are in the
transmitting (on) position. If a node loses all
channels from one side, i.e., the power coming from one
side of the node is lost, the switch on this side of the
node shuts off. Since shutting off that switch breaks
both directions, the switch in the node on the other
side of the break will also detect a power loss and shut
off. Thus, the switches on either side of a cable break
shut off if the power disappears, ensuring that a break
in just one fiber will be noticed.
For example, referring to Fig. 2d, assume that
initially switch S on the right side of node 1' and
switch S on the left side of node 2' are off to prevent
circulation of ASE. If a cable break occurs between
nodes 3' and 4', the detector D on the left side of node
3' and the detector D on the right side of node 4' will
detect the break. Signals will be sent along
connections 7 to cause the switch S on the left side of
node 3' and the switch S on the right side of node 4' to
turn off. Nodes 3' and 4' will then communicate with
nodes 1' and 2' and cause the switch S on the right side
of node 1' and the switch S on the left side of node 2'
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
14
to turn on either by one or more pilot tones or by a
separate wavelength channel.
In accordance with Applicant's invention, when a
cable break occurs, the node originally having the
blocking (off) switch resets its switch to a
transmitting (on) position and the node switch (or
switches) adjacent to the cable break shuts off, the
latter to ensure a break in both directions. In this
way, a bus is provided that can change its starting and
stopping point depending on where a cable break occurs.
A great advantage is that none of the nodes are isolated
by the cable break, and all transmitters and receivers
can remain in the same operating condition as before the
break.
Since the optical bandwidth of the network is
limited, it is desirable to keep the number of required
wavelengths to a minimum. Waveform reconfiguration
permits wavelengths to be re-used, thus reducing the
number of required wavelengths in the network. If a
reconfiguration of the wavelengths is permitted when a
cable break occurs, the required number of wavelengths
decreases from M(M-1)/2 to (M? 1)/4, where M is the
number of nodes. (Of course, if the number of nodes is
even, the required number of wavelenghts would decrease
from M(M-1)/2 to M2/4). By re-configuring wavelengths,
the number of nodes in a network can be increased
without increasing the number of required wavelengths.
This allows for large growth of the network to service a
greater number of users.
Wavelength reconfiguration can be done by utilizing
either tunable lasers or extra lasers in the nodes.
Tunable lasers permit selection of particular
wavelengths as carriers for particular respective
information, and tunable filters, such as the add-drop
filters described in more detail below, are useful as
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
wavelength-selective elements in the tunable lasers for
= selecting the particular wavelengths. At the receiving
end, using tunable filters as the selectors 22-32 in
Figure lb provides simultaneous selection of desired
5 channels by the filters and detection of the channels by
the receivers connected to the filters.
Figures 4a-4f illustrate examples of add-drop
filters for a network as described above. The add-drop
filters depicted in Figures 4a-d and Figure 4f permit
10 re-use of a wavelength by dropping information carried
on a particular wavelength and adding different
information carried on the same wavelength.
Figure 4a shows an add-drop filter that includes an
optical circulator 50 and an adjustable fiber Fabry-
15 Perot etalon (FPE) 52. A receiver R and a transmitter T
are coupled to the fiber FPE 52 by a suitable device
such as an optical splitter 54. Of course, a single
transmitter or receiver could be provided after the FPE
52. The optical circulator 50 propagates wavelength
channels from port to port in a conventional way, and
the FPE is adjusted to pass a selected one of the
wavelength channels. In Figure 4a, information X3
arriving at the node is passed through the FPE 52, which
is tuned to the wavelength X3, to the receiver R and is
replaced by departing information .3 emitted by the
transmitter T that passes through the FPE 52.
Suitable optical circulators are commercially
available from several manufacturers, including the
model CR1500 made by JDS Fitel. The Fabry-Perot etalon
may be of a commercially available type, such as the
FFP-TF made by Micron Optics. Fabry-Perot-etalon
filters exhibit high levels of stability and
performance, ease of tuning and locking onto a given
wavelength, and small, rugged packaging, which makes
them ideal for WDM applications. The above-cited paper
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
16
by C. M. Miller et al. describes this in more detail and
is incorporated here by reference.
Figure 4b illustrates an add-drop filter that uses
fiber gratings, the performance of which is described in
the above-cited paper by V. Mizrahi et al. that is
incorporated here by reference. The filter includes an
optical circulator 50, a receiver R, a transmitter T,
and a device such as an optical splitter 54 arranged in
a manner similar to the filter illustrated in Figure 4a.
Instead of the FPE 52, three fiber gratings 56, 58, 60
collectively act as a simple band-pass transmission
filter for passing the carrier wavelength X3. Each of
the gratings 56, 58, 60 rejects a wavelength, and
together they reject all channels not used in the node.
The desired wavelength X3 passes unaffected between the
grating stop-bands. In Figure 4b, arriving information
X3 passes through the fiber gratings to receiver R and
is replaced by departing information X3 that passes
through the fiber gratings and is added to the channels
Xl, X2 and X4.
Fiber gratings are simple, insensitive to
polarization, have flat pass-bands, and are compatible
with optical fibers. Fiber gratings also exhibit
outstanding cross-talk performance, which makes them
particularly suited to WDM optical communication systems
that typically require small channel spacings.
Figure 4c illustrates another embodiment of an add-
drop filter that includes two three-port FPEs 62, 64,
each of which captures optical power rejected
(reflected) from the etalon so that the third port acts
as a tunable band-stop or notch filter.
A three-port FPE is an etalon having one in-port
and one out-port on one side and one out-port or in-port
on the other side, as depicted for example in Figures 5a
and 5b. The FPE 62 in Figure 4c has two in-ports, one
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
17
on each side, as in Figure 5b. In Figure 4c, the three-
port FPE 62 is tuned to channel X3 and prevents
departing information X3 emitted by transmitter T from
passing through to the receiver R.
Figure 4d shows an add-drop filter that includes
an acousto-optic transmission filter (AOTF), one of
which is described in the above-cited paper by J.E.
Baran et al. that is incorporated here by reference.
The AOTF acts as a switch that directs nearby channels,
simultaneously and independently, to either of two
ports. The AOTF distributes channels by flipping the
polarization state of a carrier having an optical
frequency that is a multiple of an applied sound
frequency Rf. Optical channels having frequencies that
are multiples of the applied sound frequency Rf are
directed into one port of the AOTF, while other optical
channels are directed to the other AOTF's other port.
In Figure 4d, the applied sound frequency Rf is assumed
to contain a frequency corresponding to wavelength X2
and a frequency corresponding to wavelength X4 so that
arriving old information X2 and X4 is directed into the
same output port and channels X1, X3, X5 are directed to
another output port. At the same time, departing new
information X2 and X4 that is provided to another input
port of the AOTF is directed to that other output port.
The AOTF is uniquely suited for WDM applications
because it can simultaneously route many wavelength
channels, selected substantially at random; except for
the relationship between optical frequency and applied
sound frequency, there is no constraint on the order or
adjacency of the selected wavelengths. Integrated AOTFs
are not yet commercially available, although AOTFs can
be made of commercially available discrete components.
Other ways of implementing an add-drop filter can
be used, as shown in Figures 4e and 4f. The simplest
CA 02218089 1997-10-10
WO 96/32787 PCT/SE96/00441
18
way to add channels to a fiber is to use ordinary
couplers 66, 68 as shown in Figure 4e. The add-drop
filter of Figure 4e does not permit wavelength re-use.
To be able to re-use wavelengths, fiber gratings 56, 58,
60 are placed between the two couplers 66, 68, as shown
in Figure 4f. An integrated Mach-Zehnder filter could
also be used.
Re-using wavelengths requires high-performance add-
drop filters. Some of the filters described above may
need further development before they fulfill the
required demands.
Applicant's FBDNA solves the problem of protection
switching of the traffic in an optical network in an
easy way by combining the advantages of ring- and bi-
directional-bus structures. Since the FBDNA is
optically a bi-directional bus, it also solves the
problem with circulating ASE. The FBDNA requires few
and not very advanced components and is therefore a good
economical approach to communication.
While particular embodiments of Applicant's
invention have been described and illustrated, it should
be understood that the invention is not limited thereto.
This application contemplates any and all modifications
that fall within the spirit and scope of Applicant's
invention as defined by the following claims.
