Note: Descriptions are shown in the official language in which they were submitted.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
PHOTONIC COMMUNICATION SYSTEM WITH S>~J8-LINE RATE
BANDWIDTH GRANULARITY, PROTOCOL TRANSPARENCY AND
DETERMINISTIC MESH CONNECTIVITY
BACKGROUND OF THE INVENTION
Field of the Invention
The present invention relates generally to telecommunications networks that
have a
plurality of nodes interconnected by an optical transmission medium, and
particularly to
self healing optical single and multiple wavelength networks with hubbed,
meshed or
mixed connectivity. More particularly, it relates to such networks with "sub-
line rate"
bandwidth granularity, protocol transparency and deterministic mesh
connectivity.
Prior Art of the Invention
Today's telecommunications networks typically consist of access networks that
connect
end-users, also referred to herein as clients, to the network and transport
networks that
provide the interconnection between the access networks. The transport
networks can be
further separated into metro, regional inter-office facilities (IOF), also
refereed to as metro
core, and a backbone or core portions.
The access networks are under pressure to increase the variety of supported
protocols to
support emerging services, which typically require higher bit-rates, such as
private-line
Ethemet (TM). The transport network in-turn, are under pressure to provide
more capacity
and switching flexibility to support the increase in capacity coming from the
access
networks.
Optical telecommunications networks are currently the predominant architecture
for
transport networks to connect optical nodes to transfer voice, text, data,
video information
etc., referred to herein as tragic; more specifically including a variety of
optical network
topologies, such as point-to-point, linear add-drop, ring and mesh optical
networks. In the
SUBSTITUTE SHEET (RULE 26)
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
2
event of a failure re-routing of traffic towards the opposite direction is
done using spare
capacity, lower priority capacity or dedicated protection capacity.
Currently, the key standard for conventional optical networks is time division
multiplex
(TDM) based SONET/SDH (Synchronous Optical Network/Synchronous Digital
Hierarchy). SONET was developed to provide a survivable transport
infrastructure for a
wide variety of traffic protocols and bit rates. The SONET/SDH standard
defined a
hierarchy of optical transmission rates- optical carrier (0C) level for SONET
and
synchronous transport mode (STM) for SDH. For example, SONET optical Garner -
level
3 (0C-3) transmits at 155 Mb/s, while SDH synchronous transport mode level 1
(STM-1)
transmits at 155 Mb/s, over different network topologies.
In order for SONET/SDH to cant' a range of traffic protocols and bit rates,
referred to also
as payload protocols and payload bit rates, SONET/SDH defines a payload
"envelope" into
which all pre-defined SONET/SDH supported payloads must be mapped. This
envelope
comprises timeslots for the traffic information and the overhead information
to manage that
traffic. This provides SONET/SDH with the ability to carry a range of
protocols; however,
a new protocol cannot be transported until a mapping is defined so that an
interface (port)
circuits is developed, then verified, and then finally deployed. Even with
"virtual
concatenation" this approach still is the norm. If the bit-rate of the new
protocol is above
the capacity of the local network infrastructure then the entire local
network, including all
the nodes on that network may have to be upgraded.
Recently optical telecommunications networks have provided increasing capacity
using
wavelength division multiplexing (WDM), or dense wavelength division
multiplexing
(DWDM). The term "wavelength" is defined herein as an end-to-end optical
channel or
circuit of the same optical frequency from source to destination across an
optical network.
In practice wavelengths may change frequency through wavelength translation to
make
longer distance connections and/or to avoid wavelength blocking at
intermediate nodes.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
Photonic communication systems which include switching nodes which route
optical
signals without converting the signal from optical (O) to electrical (E)
signals and back to
optical (O) again (0E0 conversions) are soon to move from the lab to practical
deployment. These systems provide substantial benefits over existing systems,
in which
optical signals are switched almost exclusively in the electrical domain, i.e.
Canadian
Patent 2,271,813 , but they also have shortcomings. These systems are based on
switching
all the data in a given wavelength from one path to another, resulting in
either inefficient
transport, due to low data rates; or excessively large bandwidths being
switched. A key
impediment to more efficient processing of the bandwidth is the data
transmission format,
typically SONET or SDH, which does not lend itself to simple optical
management. An
alternative method being pursued is the use of optical packet switching, see
Canadian
Patent 2,310,856, in which, analogously with electrical packet switching,
optical packets
with associated routing information are transmitted and optical switches must
determine
the appropriate route for each packet. These systems must deal with contention
for
transmission resources at each node, require substantial effective bandwidth
for each
packet label, require extremely high speed optical switches, and require high
speed
processing at the nodes to determine the appropriate path through the node.
An improvement provided by the present invention is that the optical data
signal is
presented to the network with a format that is conducive to optical management
as a
connection with a rate less than or equal to the line rate, but in a format
which does not
require very high speed operations, pre-calculated paths simplify contention
avoidance and
is indifferent to the underlying protocol and bit-rate of the data traffic
being transported and
photonic switches are employed for fibre to fibre routing (i.e. cross-
connection), which
enable the bandwidth management without re-conversion back to electrical. The
optical
path for data through the network is established once per connection, and so
all contention
issues can be resolved in longer times or the connection can be disallowed,
without the
danger of a partial connection. The removal of both the OEO operations and the
need for
large bandwidth aggregation machines (i.e. Multi-Protocol Label Switching
(MPLS),
Asynchronous Transfer Mode (ATM) or Synchronous Transport Signal (STS) cross
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
4
connects) results in substantial savings in the capital and operating costs of
a photonic
network.
The teen "node" is defined herein as an entity comprising client ports to
receive and
transmit data from devices such as add-drop multiplexers (ADMs), routers,
switches etc.
for transporting client "data" traffic, and facility ports to deliver data
trai~c to and from
other nodes in the network. The node also may have an optional traffic control
and
management unit. The optional traffnc control and management unit has some or
all the
capabilities to transfer, multiplex and demultiplex, process, monitor, protect
(1+l; 1:1; 1:N;
where N is the number of working links or units that share the protection link
or unit),
switch or route the signals inside the node.
A "point-to-point" network is defined herein as group of entities comprising
two nodes
directly connected with no intermediate nodes and all the traffic begins and
ends at the
nodes. The physical connection is made by one or more optical fibers; called a
"span".
A fiber "span" is defined herein as a set of working and spare protection
links or capacity
in parallel between adjacent nodes, with single or multiple wavelengths.
A "linear chain network" is a point-to-point network, but with intermediate
nodes where
traffic can be dropped (received by) or added (transmitted from) at the
intermediate nodes.
A tree and branch topology is a variant of a linear chain network.
A "ring network" is defined herein as a group of entities comprised of uni- or
bi-
directionally connected nodes in a physical or logical loop with fiber spans
between any
two nodes, all nodes are 2-connectd (i.e. each node has 2-spatially diverse
routes emanating
from that node, one to an upstream node and one to a downstream node), with
single or
multiple wavelengths, and with working and protection capacity around the ring
or
between two or more nodes. In the event of a failure of one of the diverse
routes, spare
capacity on the other route is used to restore the ring traffic affected by
the failure.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
The most predominant rings in today's optical telecommunications networks are
path-
switched Unidirectional Path-Switched Ring (UPSR) or line-switched Bi-
directional Line-
Switched Ring (BLSR).
In path-switch ring traffic protection is path based. A "path" is a SONET/SDH
term for a
transport traffic connection all the way between two path-terminating
equipment (PTE)
nodes. In the event of a failure the entire path is moved (i.e. switched) over
a to a
protection path.
In line-switched ring trafl~c protection is line based. A "line" is a term for
a SONET line or
SDH multiplex section for a transport traffic connection between each pair of
line-
terminating equipment (LTE) nodes. In the event of a line failure, only that
part of the
traffic route is changed when the trafl~c is moved (i.e. switched) over to a
protection line at
the fault's boundary between the pair of LTEs.
A "mesh network" is defined herein as a group of entities comprised of three
or more uni-
or bi-directionally connected nodes, with fiber spans between any two nodes,
with nodes
that are "n-connected" where unlike rings n can be more than 2 (i.e. 3-
connected), with
single or multiple wavelengths, with working and protection capacity in the
mesh or
between two or more nodes, and with high physical connectivity.
A "constrained mesh network" is defined herein as a mesh network that has had
its
architecture constrained by practical factors such as geography, hierarchies,
commercial
restrictions etc. that limit its physical and logical mesh connectivity. A hub
and spoke
network is a variant of a constrained meshed, whereby all the n-connected
nodes are
constrained to being co-located at one or two main sites.
The conventional optical network architecture is designed with little
dependence on, or
awareness of, the connections between multiple rings or meshes, for
connectivity or
protection.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
6
The conventional optical network architecture is mufti-layer (i.e. optical
transport and
electrical multiplexing and switching) and mufti-protocol (i.e. TDM and IP).
The conventional optical network traffic is deterministic versus best effort
(i.e. Internet
Protocol networks).
The conventional optical node architecture is designed with electronic based
traffic
management and control units for managing traffic granularity.
As network traffic increases, service and cost considerations, along with
technology
advances, are driving the conventional telecommunications networks to de-layer
to fewer
layers (in order to become more scaleable), namely to an optical physical
layer and an
electrical service (i.e. packet) layer. This requires the optical physical
layer to become more
optically granular and service transparent, while still maintaining traffic
determinism if
transport Garners are to remain competitive and flexible as this de-layering
proceeds, and to
keep being a robust (i.e. 99.999% availability which corresponds to less than
S minutes of
down time per year) and scaleable transport provider for the underpinning
service layer.
The optical granularity increases flexibility and maximizes bandwidth
efficiency to keep
the carrier's optical bandwidth cost competitive so that the Garner can keep
providing
traffic transport and traffic restoration for the layer above.
SUMMARY OF THE INVENTION
The present invention endeavors to provide an improvement over existing
optical
communication systems of the optical transport of granular capacity of
bandwidths less
than the line rate capacity ofthe optical transport on a given optical
frequency, sometimes
referred to as the sub-wavelength or sub-lambda level, with full optical
connection oriented
bandwidth management, including but not limited to, connection establishment,
re-
arrangement, protection, route diversity, restoration, aggregation,
separation, switching and
mufti-cast, of that granular capacity, which alleviates totally or in part the
drawbacks of
prior art, such as SONET/SDH based networks.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
7
According to the present invention there is a provided an optical
communications network
employing wavelength division multiplexing with wavelength hopping and TDM
bursts,
comprising a plurality of nodes; aggregation nodes and switch nodes; a
transmission
medium interconnecting said nodes, said transmission medium being capable of
carrying a
plurality of wavelengths that are capable of being shared with other optical
communications networks; and an interface at each node for dropping a
wavelength
hopping optical TDM burst for a controlled interval therewith, adding a
wavelength
hopping optical TDM burst for a controlled interval destined for another node,
and
passively or actively, the latter for signal conditioning purposes, forwarding
wavelength
hopping optical TDM burst for a defined interval destined for other nodes; and
whereby an
interface at each switch node for cross-connecting or switching or overlaying
wavelength
hopping optical TDM burst between a plurality vftransmission medium; and
whereby
communication can be established directly between a pair of nodes employing
wavelength
hopping optical TDM burst without the active intervention of any intermediate
or
intervening node.; and a mechanism for maintaining optical power balance and
optical
signal integrity in the network when the wavelength hopping optical TDM bursts
are
intermittent.
It is an aspect of the invention to provide aggregation nodes with interface
devices for use
in an optical network employing wavelength hopping optical TDM bursted
waveslots,
comprising a wavelength fixed, semi-agile or fully agile de-multiplexer for
dropping
waveslots from the network at a node, means for converting the optical signal
from said de-
multiplexer to signals for generating optical or electrical output signals to
subtending
"client" devices, and a wavelength agile multiplexer for adding waveslots from
the
subtending client device optical or electrical input signal to the network.;
the said de-
multiplexer and multiplexer being arranged so as to have access if desired to
all the optical
signals. The latter in one embodiment permits inclusion of a waveslot
wavelength to
wavelength conversion and /or translation device for waveslot cross-connection
or
waveslot interchange in time, analogous to time-slot interchange (TSI). For
example, if a
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
8
connection path is established between node A and node B, over a fiber, and
between node
B and E over another fiber, but no path fiber path exists between node A and
node E, node
A can send traffc for node E first to node B, which drops the trai~c in the
form of
waveslots, detects and confirms the waveslots for node E, converts or
translates or
interchanges the waveslots through an appropriate device and forwards the
traffic onto the
fiber to node E.
A network in accordance to the invention is protocol and bit rate transparent,
where the
waveslot format is indifferent to the underlying protocol, and is therefore
more compatible
and forward evolvable with the DWDM metro transport networks that are protocol
and bit
rate independent (patent CA 02245403). Each trai~c payload is carried on
separate
protocol and bit rate transparent wavelength hopping optical TDM bursts that
can be
aggregated, separated or rearranged amongst a plurality of optical
transmission medium
using optical burst wavelength division multiplexing techniques.
An aspect of the invention is that cascaded rings can be supported with inter-
connecting
nodes as per patent CA 02245403, but since the network can be synchronized to
an external
synchronization source, such as to a carrier's building integrated timing
supply (BITS),
non-linear effects such as chromatic dispersion accumulation, and spatial
effects like fitter
accumulation, can be mitigated without the regeneration complexity as stated
in patent CA
02245403.
An aspect of the invention is that optical gain blocks such as fiber
amplifiers, such a
erbium doped fiber amplifiers (EDFAs), or specialized short fiber amplifiers,
or silicon
optical amplifiers (SOAs), and linear optical amplifiers (LOAs) are supported
for adding
amplification to individual wavelengths or groups of wavelengths to achieve
the required
optical system bit error rate performance.
According to the invention a waveslot is associated with the connection
between two or
more nodes without the need for the nodes to be on the same pre-assigned
"band" of
wavelengths as per patent CA 02245403.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
9
In another aspect, the present invention provides a line of sight connection
function, with
signaling, protection and restoration functions, for the transport of granular
optical capacity
of bandwidths less than the line rate capacity of the optical transport,
refereed to hereinafter
as a Line of Sight Connection Protocol (LOSP). The LOSP for use with the
described
signaling format and switching method to enable connection oriented bandwidth
management at the granular level, such as the sub-wavelength level, in a
network
optimized approach, to provide an additional increase in network bandwidth
e~ciency and
flexibility.
According to this aspect of the present invention, the end-nodes of a
connection to be
established perform the connection establishment and /or re-arrangement
process using the
LOSP.
According to a preferred aspect the invention, the end-nodes of a connection
establish the
protection for the connection (whether by dedicated redundancy using a
previously
unassigned connection or spare link, shared redundancy using a lower priority
connection ,
optical route diversity, or inter-layer route diversity) using the said LOSP.
The end-nodes of a failed shared protected connection to be restored perform
the
preemption process on lower priority connections to restore on-demand the
higher priority
shared protected connection using the LOSP. The pre-empted path may be pre-
configured
or may be determined at the time of the fault event. The LOSP can perform the
restoration
in the optical layer or in co-ordination with a higher network layer (such as
the Internet
Protocol (IP) layer 2/3 levels) in the routers for example.
To be re-established; the end-nodes of a suspended lower priority connection,
suspended
say for the purposes of immediate restoration of a shared protected
connection, perform the
connection restoration process using the LOSP.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
The present invention also provides a shareable network-wide, optimized,
granular
connection capacity, based on information stored and provided by each node,
that is
coordinated at the network level utilizing LOSP to provide all the above
described
functions optimized at the optical network level, including any required inter-
layer co-
ordination for connection management, aggregation, pre-emption, route
diversity.
An advantage of the present invention is improved optical bandwidth
efficiency.
The invention provides a flexible method for granular bandwidth management, in
a wide
variety of optical network topologies, including, but not lunited to, point-to-
point, linear
10 add-drop, collector daisy chain, ring and mesh, with a plurality of
connection, protection
and restoration options.
BR1EF DESCRIPTION OF THE DRAWINGS
The preferred embodiments of the invention will now be described in
conjunction with the
annexed drawings, in which:
Figure 1 shows a schematic of a reference transport network illustrating the
physical
layout of a typical transport telecommunication network within which the
present invention
is applied;
Figure Z shows a block diagram schematic of a granular optical burst network
showing the
physical layout of an optical burst, a multiplexed and switched network
according to the
presentinvention;
Figure 3 illustrates operation of an optical burst network in an example of a
mesh
connection pattern on an optical burst network as shown in Figure l with
examples of
payload signals in the waveslot connections, according to the present
invention;
Figure 4 illustrates the optical burst data format in accordance with the
invention;
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
1. l
Figure 5 illustrates the sequencing of the cycles optical burst waveslot of
the data format
shown in Figure 4;
Figure 6 illustrates agile bandwidth mapping into waveslots by optical burst
switching of
the waveslots from one to two individual transmission mediums with a 3-
connected
granular optical switch, in accordance with the invention;
Figure 7 illustrates the agile mapping of the bandwidth into wavelength
timeslots, referred
to as "waveslots" herein, for a typical system in accordance with the
invention;
Figure 8 illustrates bandwidth aggregation of waveslots, in this case from
four
transmission mediums to one;
Figure 9 illustrates how SONET/SDH rates (OC-3/STM-1, OC-12/STM-4 and OC-
192/STM-64) are granularized with the waveslot format;
Figure 10 illustrates the compatibility of the waveslot photonic layer format
of Figure S for
carving a variety of data of various protocol and bit rate payloads;
Figure 11 is a system block diagram of a network granular aggregation node
with optical
burst multiplexing and optional switching capability;
Figure 12 is a functional block diagram of an example implementation of the
two or four
fibre network node of figure 11;
Figure 13 is a functional block diagram of an example implementation of the
four fibre
network node of figure 11;
Figure 14 is a system diagram of a network granular switching node, in this
case eight-
connected, with optical burst switching capability;
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
12
Figure 15 is a more detailed system diagram of a network granular switch node,
with a
folded plane and optical burst switching capability;
Figure 16 is a yet more detailed rendition of the system diagram of figure 15;
Figure 17 illustrates the main options for granular aggregation node to switch
node
connection via line facility;
l 0 Figure 18 illustrates granular aggregation node to switch node connection
via direct
connection to the switch fabric;
Figure 19 illustrates the main option for aggregation node to aggregation node
connection;
Figure 20 illustrates a Less service-disruptive and lower pass-through loss
interconnection
option for expandable aggregation node to aggregation node connection;
Figure 21 is a functional block diagram of an example implementation of
waveslot
alignment;
Figure 22 illustrates an example of a LOSP operation;
Figure 23 illustrates an example of a LOSP operation to request the
establishment (setup)
of a connection path or route;
Figure 24 illustrates the functional format of a LOSP connection seeking
control packet
(CSP) or message;
Figure 25 illustrates an example of a LOSP operation to identify open channels
for an
optimal connection path or route;
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
13
Figure 26 illustrates an example of a LOSP operation for reserving a
connection path or
route;
Figure 27 illustrates the functional format of a LOSP connection reservation
request
control packet or message; and
Figure 28 illustrates an example of a LOSP operation when the desired
connection route is
blocked.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
Referring to figure 1, it shows a typical telecommunications network
consisting of access
networks that connect end-users, also referred to herein as clients, to the
network and
transport networks that provide the interconnection between the access
networks. The
transport networks can be further separated into a Metro, Regional Inter-
Office Facilities
(IOF), also referred to as Metro Core, and a backbone or core portions. In
Figure l, blocks
109,110,112 are metro hub sites of the metro portion of the transport network
blocks 114,
117,124 are regional hub sites of the regional portion of the transport
network and block
105 the backbone portion of the transport network. Thus, Figure 1 shows the
physical
layout of the network. The access networks are under pressure to increase the
variety of
supported protocols to support emerging services, which typically require
higher bit-rates,
such as private-Iine Ethernet (T'"). The transport networks in-turn, are under
pressure to
provide more capacity to support the increase in capacity coming from the
access networks.
Figure 2 shows an embodiment of the present invention wherein a plurality of
aggregation
nodes 132,135,137,139, 143,144 and 148 are provided for aggregating and
separating the
optical data traffic and switch nodes 131, 138, 142, 153 and 156 for cross-
connecting the
optical data traffic, interconnected in an arbitrary network topology,
including rings (as in
nodes 132, 134,135,137,142), meshes (as in nodes 131,142,138,153,156) and
linear chains
(as in nodes 143 and 144), by optical transmission media 134,157,150 capable
ofcarrying
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
14
a plurality of wavelengths. It will be understood that Figure 2 shows the
physical layout of
the network. The interconnectivity between the nodes is provided by the
wavelength
hopping optical bursted waveslots. A device, either wavelength fined or agile
is provided at
each aggregation node 1.32,135,137,139, 143,144,148 for dropping and adding
the
associated wavelengths for a given time interval, wavelength hopping optical
bursted data
unit (ODU), as fixed length frames in time slots, called waveslots herein,
within a repeat
interval, and passively forwards other waveslots designated for successive
nodes over the
transmission medium.
For the ring, on medium 134 each node 132, I35, 137, 142 can add/drop
waveslots specifrc
to that node. In order to establish a connection say between node 132 and 137,
node 132
transmits in both directions for a 1+1 or I:l protected connection, on the
counter rotating
rings of 133 and I 36 waveslots for node 137. 133 and 136 provide two diverse
routes on
the ring. In the event of a failure of one ring arc, say 136, the other ring
arc, in this case 137
provides a restoration path for all the waveslots from the now failed arc 136.
The waveslot
on ring arc 136 passes through node 135 where it is either passively reflected
or actively
passed or conditioned and forwarded to node 137 that drops the waveslot and
extracts the
traffic in the waveslot payload. The waveslot on ring arc 133 passes through
switch node
142 where it is either passively reflected or actively passed or conditioned
and forwarded to
node 137 that drops the waveslot and extracts the traffic in the waveslot
payload. In
accordance to the principles of the invention, the waveslots that permit the
direct, protocol
transparent and independent connections to be made between any nodes on the
ring without
the intervention of any intermediate node. The nodes on the ring can be
logically
interconnected in various connection manner, for example hubbed, star, meshed
etc. by
establishing the appropriate connections between the nodes on the ring. If
connected in
rings, these rings may be connected together such that data tragic can be
transmitted and
received between adjacent rings.
For the mesh, on media 154 and 155, each switch node 131, 142, I38, 153,156
can re-
arrange overlapping waveslots between the plurality of optical medium going
into (inlet or
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
1S
incoming or ingress or connected from) and out of (outlet or outgoing or
egress or
connected to) that node. In order to establish a connection say between node
13S and 139,
node 13S transmits in both directions, on the counter rotating rings of 133
and 136
waveslots for node 139. The waveslot on 136 passes through node 137 where it
is either
passively reflected or actively passed or conditioned and forwarded to switch
node 14Z that
per its waveslot connection map, redirects or "switches" the waveslot to a
outlet fibre that
will carry the waveslot to node 139. In this example, assume that is the fiber
to switch node
138. The waveslot goes to switch node 138 where it is redirected to a fiber
that in this case
connects directly to aggregation node 139, its intended destination. Node 139
drops the
waveslot and extracts the traf-1-~c in the waveslot payload. The waveslot on
133 passes
through node l 32 where it is either passively reflected or actively passed or
conditioned
and forwarded to switch node 142 that per its waveslot connection map,
redirects or
"switches" the waveslot to a outlet fibre that will carry the waveslot to node
139. In this
example, assume that is the fiber to switch node 156. The waveslot goes to
switch node 1 S6
where it is redirected to a fiber to switch node 1 S3, which re-directs it to
a fiber to switch
node 138 that in this case connects directly to aggregation node 139, its
intended
destination. Node 139 drops the waveslot and extracts the traffic in the
waveslot payload.
In accordance to the principles of the invention, the waveslots that permit
the direct,
protocol transparent and independent connections to be made between any nodes
on the
mesh using a format that permits granular all-optical switching of the
waveslot at the
switch node from a plurality of incoming fibers to a plurality of outgoing
fibers. The nodes
on and connected to the mesh can be logically interconnected in various
connection
manners, for example hubbed, star, meshed etc. by establishing the appropriate
connections
between the nodes on the mesh.
For the linear add-drop chain, a variant of an optical tree, on physically
diversely routed
medium 1 S 1 a, 1 S2a and 1 S 1 b, 1 S2b, each node 143, 144, l S3 can add
drop waveslots
specific to that node. In order to establish a connection say between node 143
and 148,
node 143 transmits on l S2a, and if the connection is 1 +1 or 1:1 protected on
1 S2b for node
148. The waveslot on 1 S2a passes to node 144 where it is dropped for
forwarding to node
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
16
148, over 149 that drops the waveslot and extracts the traffic in the waveslot
payload.
Likewise the waveslot on 152b passes to node 144 where it is dropped for
forwarding, over
149, to node 148 that drops the waveslot and extracts the traffic in the
waveslot payload.
Referring now to figures 3 and 4, it is the waveslot format that allows the
network to be
protocol independent. A device, either wavelength fixed or agile is provided
at each
aggregation node A of the aggregation node chains 186,187,164,175,174 and 173
for
adding waveslots from the node to the transmission medium connected to the
switch nodes
S at the head of each chain. Communication betweens nodes, such as between l
86 and 173
for the purposes of management and control can be established directly with a
dedicated
management waveslot or indirectly by appending the information to a traffic
carrying
waveslot between the pair of nodes 186 and 173 without the active intervention
of
intermediate nodes.
The device, either wavelength fixed or agile, provided at each aggregation
node A of the
aggregation node chains 186,187,164,175,174 and 173 for dropping and adding
the
waveslot can be programmed so that the dropped and added waveslots are at
dii~erent
wavelengths for a connection. This permits lower optical isolation variants of
components,
such as optical filters, to be used for cost sensitive applications.
Each transmitter can provide a single colour at any given time, so the link
state-matrix is
'singly filled'. The optical switch nodes are able to route the waveslots
through the nodes,
and so overlay the link state matrices 159, 176,184 to make multiply filled
matrices, such
as 160, with each wavesIot entering the switch from an input (inlet or ingress
or connected
from) fibre following the required path through the node and onto the
appropriate output
(outlet or egress or connected to) fibre.
A connection from a transmitter to a receiver is formed as an optical signal,
which is
transmitted in the correct waveslot (correct timeslot and at the desired
wavelength) to
traverse the switch node where it can be switched from the inlet or ingress
optical fibre to
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
17
the egress or outlet optical fibre as it moves from switch node to switch node
in the
network. The path through the network the data signal traverses is controlled
in this manner
by the optical switches through which it propagates until it reaches its
destination. The
choice of waveslot (wavelength and timeslot) for transmission is determined by
the
network connection setup system and connection setup protocol.
Figure 3 shows a mesh connected network, ring structure shown for clarity. The
switch
nodes S are photonic cross-connects that space switch like-waveslots per
wavelength plane.
Switch node 170 manages tandem traf~rc without optical add/drop. Aggregation
node 164
is providing OC-192 pass-through on the second wavelength as indicated in
pattern 159.
Node 186 is transporting GbE tragic as indicated in pattern 185.
The waveslot 189 in Figure 4 is the fundamental unit of bandwidth for the
invention. The
waveslot is connection oriented. The waveslot is used to transport data trat~c
in fixed or
variable intervals in duration on wavelengths on the transmission medium
between nodes,
which may be non-adjacent. The same waveslot in each cycle has the same
wavelength to
simplify connection management. Higher bandwidths than the capacity of a
waveslot per
connection channel are achieved by using multiple contiguous or non-contiguous
connections. Having the waveslots contiguous simplifies alignment for
switching. The
waveslots are transmitted in a repeating cycle, with typically a fixed
duration on each
wavelength. The cycle time is chosen based on update rate and desired latency
through the
network. The waveslot duration is chosen for optimum bandwidth granularity -
minimum
managed bandwidth = (line rate)/(number of waveslots per cycle).
For example, sixty-four forty-microsecond waveslots will fit within a 2.56 ms
repeat
interval. If the nominal line rate is ~10 Gb/s, then each waveslot within the
repeat interval
equates to a connection of 150 Mb/s
The format of the waveslot is optical transport Network (OTN) compatible.
Current ITU-T
OTN defines multiple channels per wavelength, with nominally one channel per
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
18
wavelength. The channels supporting path rates of 2.SGbps, lOGbps, or 40 Gbps.
The
channels have a common digital frame structure, with defined payload and
overhead
information area. The invention supplements the OTN compatible format with the
concept
of a sub-wavelength channel that has a repeating OTU frame compatible size.
425 OTN
waveslots per second provides a nominal SOMbps Sub-wavelength channel.
Therefore a
lOGbps wavelength will have 192 sub-wavelength channels that can be
transported in 192
waveslots per cycle 194 in Figure 4.
The connection capacity of a switch equals total number of channels equals
SxFxC,
where s = number of waveslots, F = number of fibers, and C = number of colours
(wavelengths or lambdas). For S=16 @ OC-12, C=40, F=6 the total number of OCl
2
channels equals 3,840. Where S=64 @ OC-3, C=40, F=6 the total number of OC-3
channels equals 15360.
The waveslot format shown in figure 4 is referred to herein also as the
Photonic burst
Switch Transport format (PSTF). Waveslot 189 consists of the preamble 188 of
bits that
serves as a label or tag to identify the source of the waveslot and contains
information that
a receiver can use to verify there is no collision with another waveslot or
wavelength from
another transmitter, the payload 190 that carnes the data traffic, and assists
in clock
recovery and the tail label or tag, 191, that contains error detection and
connection
management information. Aside from the clock recovery assist information the
rest of the
information in the head and tail tags does not have to be ahead of the data
payload 190, and
could even be superimposed on the payload, say by using a sub-carrier or a
similar
technique.
A connection is a waveslot, channel rate (i.e. the service rate, like OC-
12/STM-4 where 4
concatenated waveslots are required per connection), colour (i.e. wavelength),
and fibre
combination available through the system from source to destination. The
connection is
typically bi-directional and symmetrical, but can be uni-directional,
asymmetrical, diverse
paths for each direction etc.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
19
The transmitter at the node, whether fixed or agile provides the dynamic
wavelength
allocation, called wavelength hopping herein, for each waveslot. For maximum
system
flexibility the transmitter and receiver at the node has access to all the
operational system
wavelengths.
The information contained in 191 and 188 is employed by the connection
management
system to detect erroneous connection states such as misconnection, multiple
connections,
multiple transmitters (senders) and multiple receivers (listeners).
The duration 195 of a waveslot is flexible, but will be typically fixed for
the network.
Larger transport bandwidths can be achieved through the use of multiple
contiguous or
non-contiguous waveslots or wavelength. 192, the blank or undefined interval,
delineate
and is also a guard or transition band for a waveslot. The blank, or undefined
interval may
be populated with fill bits, training bits, marshalling bits or the like to
speedup the burst
receiver's acquisition times. Each waveslot in a cycle 194 may have a
different
wavelength.
The same waveslot, for example l 93, in each cycle has the same wavelength.
Waveslots
are routed as connections. Optical switches capable of switching at the
waveslot level
overlay cycle 194. Optical switch transition occurs during the blank/undefined
time 192.
Adjustment of optical fiber to fiber timing alignment occurs during the
blank/undefined
time 192. The management system ensures there are no data collisions amongst
waveslots
on the transmission medium.
Figure 5 shows the wavelength hopping pattern for a single agile transmitter,
in what is
refereed herein as a photonic link state matrix, corresponding to an optical
burst switching
system with 12 wavelengths on the transmission medium. The waveslots are
represented as
squares on the pattern. Time is the vertical columns, and wavelength, 197, the
horizontal
rows, with one row per wavelength. In this case 12 rows for 12 wavelengths. An
empty
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
square, like 199, indicates an absence of a waveslot, i.e. idle or no-connect,
for that
wavelength and time, during that cycle 196. A filled square, like 200,
indicates the
presence of a waveslot, i.e. busy, for that wavelength and time, during that
cycle 196. The
minimum rate connection rate is one waveslot per cycle, as shown by 198, that
repeats
every cycle 196.
Figure 6 shows wavelength agile bandwidth mappings for four separate single
agile
transmitters. The photonic link state matrix has wavelength as the rows, 204.
In this
example 12 wavelengths counting from the bottom to top, and time, 206 as the
columns. A
10 single transmitter transmits a single wavelength in a single waveslot, with
the transmission
wavelength varying from waveslot to waveslot, i.e. 201 to 202, within the set
of discrete
wavelength colours used in the particular network. Typically the network may
carry, for
example, 20 or 40 wavelengths that are hopped across for providing waveslot
interconnectivity between nodes. The wavelengths can be different spacing
frequencies, for
example, 200 GHz or 1.6nm spacing, 100 GHz or 0.8nm spacing, 50 GHz or 0.4nm.
spacing etc.
In figure 6, 203 is for transmitter number one transmitting mixed services
over waveslots.
205 is for transmitter number two transmitting OC-48 based services over
waveslots. 208 is
20 for transmitter number three transmitting mixed services over waveslots.
207 is for
transmitter number four for transmitting OCl 92 services over waveslots in the
second
wavelength of the 12 wavelengths.
In figure 7 the optical switch is used for fibre-to-fibre routing of waveslots
through the
network. Each fibre supports multiple wavelengths and each wavelength supports
multiple
waveslots. Shown are the three dimensions to the bandwidth aggregation and
separation
and management. The transmitter color domain, that does the lambda hopping,
aggregation
node or time multiplexer that works in the time domain, i.e. different
waveslot bursts, and
the optical switch that works in the space domain, i.e. different fibres. The
management
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
21
system provides the connection establishment i.e. setup and connection release
i.e. tear
down.
Figure 8 shows how waveslots from four separate transmitters are aggregated
using PSTF.
The singly filled link state matrixes 222,223,224, and 225 can be aggregated
into one
combined, multiply filled link state matrix 226 as shown, at the aggregarion
node or the
switch node. The management system utilizes the LOSP protocol to enable this
aggregation
by ensuring that when signals from multiple transmitters are to be multiplexed
onto one
transmission medium, the colours of the wavelengths selected per waveslot at a
given time
by each transmitter are not overlapping with other transmitters.
Figure 9 shows the granularity maps for three transport capacities
demonstrating the
improvement in granularity with increasing waveslots per wavelength. 206 is
time, 227 is
wavelength, in this case 1 to 80. 230 is 80 channels, 1 waveslot per
wavelength, of OC-
192/STM-64 or l OGBE, ofwavelength managed services. 229 is 1280 channels, 16
waveslots per wavelength, of OC-12/STM-4 rate or GBE/2 (i.e. 640 GbE) level
managed
services, and 228 is 5120 channels, 64 waveslots per wavelength, of OC-3/STM-1
or
GbE/8 (i.e. 640 GbE) level managed services.
Figure 10 shows how the system could be run in parallel or even overlayed with
other
existing and future wavelength systems in an optical network. 231 is a link
state matrix for
an 80 wavelength optical burst network employing waveslots only for full
granular
managed bandwidth services. The network could be shared with other optical
networks by
allocating a contiguous block of wavelengths 233 to the other network for
conventional
wavelength services and leaving the rest 232 for the optical burst network for
granular
managed bandwidth services. The network could be shared with other optical
networks by
interleaving wavelengths 234 between the networks for service separation.
A typical aggregation node is shown in figure 11. 247 and 246 are the incoming
fiber (inlet
or ingress or connected from) and 239 and 240 the outgoing (outlet or egess or
connected
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
22
to) that are connected to the linear chain, ring or mesh network. 248 and 238
are optical
fiber switches for traditional tnznk protection switching, where one fiber is
selected as
'working", say 247 for unit 248, and 239 for unit 238, and the other fiber is
designated for
"protection", in this case 246 for unit 248, and 240 for unit 238. In the
event of a failure the
optical switch under the node management system supervision switches the
physical
connection from the working to the protection. 238 and 248 can operate
independently of
each other. The trunk protection switch units 248 and 238 are optional, and if
not
equipped, only a single incoming fiber connects directly to 245 and a single
outgoing fiber
connects to 237. A de-multiplexer (DEMUR) 250 connects to 248 and a
multiplexer
(MUX) 236 connects to 237. De-multiplexer 250 drops or forwards waveslots or
wavelengths to the interface units 242, 255 and multiplexer 236 adds or
forwards waveslots
or wavelengths from the interface units 242, 255.
Physically DEMUR 250 consists of either an agile optical filter or a fixed
optical filter
followed by an optical switch. The filters transmit the waveslot at the
desired wavelength
to be dropped to the interface units 242 and 255 over connections 244, and
pass the
remaining waveslots over connection 251.The agile filter can be, for example a
tunable
Lithium Niobate (LiNb03) based periodic poled filter, or a tunable Fabry-Perot
filter. A
suitable filter is in the process of being made by Dense Optics inc. of
Quebec, Canada. The
fixed filter can be, for example, interference filter, array Waveguide, fiber
Bragg Grating,
Dispersive filteretc. A suitable fixed filter is made by JDSU of Ottawa
Canada. Both types
of filters have suitable isolation, add/drop loss and pass- through insertion
loss. The
DEMUR unit can also be based on coarse filters; in that case multiple
wavelengths and
thus waveslots will be dropped referred to as gang~ropped or group dropped,
herein.
Optical switch can be a LiNb03 based switch or a Silicon Optical Amplifier
(SOA) based
optical switch. A suitable switch is in the process of being offered by
Trellis, LightCross,
Corning, and JDSU etc and is representative of other vendor's switches.
Physically MUX 236 consists of either an agile optical filter or an optical
switch followed
by a fixed optical filter or a broadband optical combiner. The filters
transmit the waveslot
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
23
at the desired wavelength to be added from the interface units 242 and 255
over
connections 241, and passed with the remaining waveslots from 250, via 25I,
252, through
an optional optical signal conditioning (and/or amplification and/or
wavelength translation
and/or wavelength conversion) unit 253 through 254 and over connection 256.The
agile
filter can be, for example a tunable LiNb03 based periodic poled filter, or a
tunable Fabry
Perot filter. A suitable filter is in the process of being made by Dense
Optics inc. of
Quebec, Canada. The fixed filter can be, for example, interference filter,
array Waveguide,
fiber Bragg Grating, Dispersive filter etc. A suitable fixed filter is made by
JDSU of
Ottawa Canada. Both types of filters have suitable isolation, add/drop loss
and pass-
through insertion loss. The MUX unit can also be based on a coarse filter; in
that case
multiple wavelengths and thus waveslots will be added referred to as gang-
dropped or
group added, herein. Optical switch can be a LiNb03 based switch or a Silicon
Optical
Amplifier (SOA) based optical switch. A suitable switch is in the process of
being made by
Trellis and by LightCross, and is representative of other vendors. The MUX can
also be
based on a broadband optical combiner. A suitable combiner is made by JDSU of
Ottawa
Canada.
The DEMI1X can be equipped with waveslot and wavelength monitoring circuitry
to
monitor optical signal performance and integrity such as optical power levels,
wavelength
accuracy, optical power stability, optical signal noise ratio etc. for both a
quality measure
and for triggering protection switching, on consistent or intermittent
degradations or faults,
or waveslot connection re-routing from degraded paths. The monitoring
information can
also be used by the optional signal-conditioning unit 253 to optimize its
operation, using
locally and/or globally driven optimization algorithms.
The dropped waveslot from the DEMUR 250 is passed to the appropriate interface
units
242 and 255.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
24
The interface unit 242 and 255 optical receive side connected to 244 from the
DEMUR 250
consists of an optical detector, burst receiver, ancillary receive electronics
to route the
signal to the client transmit circuitry connected to 243 for 242 and 235 for
255.
The interface unit 242 and 255 optical transmit side connected to 241 to the
MUX 236
consists of ancillary transmit electronics to take the signal from the client
receive circuitry
connected to 243 for 242 and 235 for 255, and pass it to mapping circuitry to
format the
signal into waveslots for forwarding to a wavelength agile optical
transmitter, such as a
tunable Iaser module, that transmits the waveslot to the MUX 236 over 241. The
laser
module ancillary circuitry with specialized electronics controls the
wavelength control,
laser current, modulation current, operating temperature if a thermal electric
cooler is
utilized for the desired wavelength, average optical power, peak power, noise
interference
compensation, extinction ratio, optical power broadband modulation etc for the
waveslot or
wavelength.
Protection links 249 exist for supporting equipment protection between 242 and
255
interface units in 1+1, l:l, 1:N etc.
The adding ofwaveslots works in the same way as dropping waveslots but in
reverse.
The nodal management system 269a and redundant units 269b for the aggregation
node
consists of a maintenance unit, a supervisory unit, an external
synchronization unit to
synchronize to BITS and an optical supervisory channel (OSC) unit. All of
which can be
1+1 or 1:1 protected. Each unit will typically contain an embedded processor
module with
processor, volatile and non-volatile RAM and ROM memory, running a mufti-
tasking
operating system. Typically FLASH memory for program store and application
store. The
units will typically have a plurality of serial and parallel, electrical and
optical interfaces
for machine-to-machine and machine-to- person communications. The application
store
contains application software for control, maintenance, status monitoring,
performance
monitoring, and network management protocols. Communication protocols,
alarming
detection and reporting, protection and restoration, communications and
control loops for
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
managing the detectors and transmitters etc. The aggregation node may be
controlled and
monitored by a software running on a remote computer, say using a telnet
session over
TCP/IP, or by a network operating system.
The aggregation node as shown in figure 12, is a more detailed representation
of the two
and four fiber aggregation node in figure 11.
The aggregation node as shown in figure 13 is a four fiber connected variant
of the node
shown in figure 12.
In figure 13 the aggregation node is equipped for four fibres. One fibre
connected to outlet
239 going to the "west" adjacent node, one fibre connected to the inlet 246
coming from
the "west" adjacent node, one fiber connected to inlet 247 coming from the
"east" adjacent
node, and one fiber connected to outlet 240 going to the adjacent "east" node.
275 and 236
are MUX units, 272 and 250 are DEMUX units, 253a, 253b are the conditioning
and/or
signal conversion and/or wavelength translation units, and/or interchange
units. 268, 264,
262,242 are the facility interface units that waveslots are dropped and added
from and to
the network, while 267,265,260 are the client interface units that interface
to the subtending
equipment whose data is being transported. Facility interface units can be
1+1, 1:1 and l :N
protected, using signals links between the units such as 249 and 261. Unit 242
is a
specialized unit that is a combined facility unit and client interface unit.
Unit 242 has also
an integrated trunk switch for protecting the client connections 243.
The nodal management system 269a and redundant units 269b is as previously
described.
The switch node in figure 14 is shown as an example. The optical taps for
monitoring are
not shown for clarity purposes. The inlet fibers 291 connect to 290 the
waveslot alignment
units, further detailed in figure 21, that compensate in the skew of the
waveslot cycles in
time due to the different fiber lengths being traversed. The appropriate delay
is switched in
for each fiber under the nodal management system control in relation to the
external
network BITS clock 285 received at the management units 284a and 284b. The
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
26
compensated optical signals are forwarded to the de-multiplexer units (DEMUR)
289 over
292.
The DEMUR can be agile or fixed filter, but fixed filter are sui~cient, for
example, 20 or
40 or 80 channel Array Waveguide (AWG) or fiber Bragg Grating or Dispersive
filter etc.
A suitable fixed filter is made by JDSU of Ottawa Canada. Any high channel
count optical
filter that has suitable isolation, little polarization dependence, add/drop
loss and pass-
through insertion Ioss can be used in DEMUR. The DEMUR unit can also be
coarse, in
that case multiple wavelengths and thus waveslots will be dropped referred to
as gang-
dropped or group dropped, herein. The DEMUR can be equipped with waveslot and
wavelength monitoring circuitry to monitor optical signal performance and
integrity such
as optical power levels, wavelength accuracy, optical power stability, optical
signal noise
ratio etc. for both a quality measure and for triggering protection switching,
on consistent
or intermittent degradations or faults, or waveslot connection re-routing from
degraded
paths. The monitoring information can also be used by the optional signal
conditioning unit
253 figure 12 and 253a and 253b in figure 13, in the aggregation nodes to
optimize its
operation, or locally in the switch node units, using locally and/or globally
driven
optimization algorithms.
The dropped waveslots based on wavelength from the DEMUR 289 is passed to the
appropriate switch units 278. The switch units can be built around Lithium
Niobate
(LiNb03) based switches or a Silicon Optical Amplifier (SOA) based optical
switch or any
sub-l OOns optical switching device. A suitable switch is in the process of
being made by
Trellis and by LightCross, and is representative of other vendors. The
switches can be as a
minimum 4x4, but can be 8x8, 16x16 etc. The switch unit is a plane space
switch that
switches the waveslot based on a connection map (i.e. look-up table) stored in
memory in
the nodal management units 284 and 284, to the appropriate multiplexer unit
(MUX) 281.
The connection map in memory is configured by a call set-up procedure which
creates the
appropriate mapping of waveslots from ingress space switch ports to egress
space switch
ports. The Iookup table validates that a waveslot corresponding to a given
switch-port and
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
27
fiber has originated from the correct aggregation node port. If the connection
management
system finds that data in a given waveslot has originated unexpectedly from an
incorrect
aggregation node port, fault correction procedures will be triggered; at the
same time the
offending waveslot will not be switched through the Photonic cross-connect
switch module
278.
The MUX unit 280 can be based on a broadband optical combiner or it can be an
agile or
fixed filter, but fixed filter are sufficient, for example, 20 or 40 or 80
channel array
Waveguide or fiber Bragg Grating or Dispersive filter etc. A suitable fixed
filter is made by
JDSU of Ottawa Canada. Any high channel count optical filter that has suitable
isolation,
little polarization dependence, add/drop loss and pass- through insertion loss
can be used in
DEMUX. The MUX unit can also be coarse, in that case multiple wavelengths and
thus
waveslots will be added referred to as gang-added or group added, herein. The
output of
the MUX unit 281 connects to the outlet fibers 280. One outlet fiber per MUX
unit. The
MUX units 281 can be optionally fitted with gain elements such as an SOA or
LOA with a
programmable attenuator. Suitable components are available from JDSU or
Corning or
Kamelian. The attenuator is used for optical power equalization amongst
waveslots passing
through 281, balancing the optical power levels from waveslots that have
traverse different
fibre distances etc therefore overcoming the problem of waveslot optical gain
adjustment.
This can be done independently of other nodes in the network or in conjunction
with them
to achieve the most optimal end-to-end system performance. The attenuators can
also be
placed in 278 either before or after the optical switch (pre- or post).
The MUX unit 280 also has tunable laser transmitters for performing optical
power fill by
inserting appropriately place optical signals in the optical spectrum for
stabilizing optical
amplifiers.
The management units 284a and redundant unit 284b for the switch node consists
of a
maintenance unit, a supervisory unit, an external synchronization unit to
synchronize to
BITS 285 and an optical supervisory channel (OSC) unit. All of which can be
1+1 or 1: l
protected. Each unit will typically contain an embedded processor module with
processor,
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
28
volatile and non-volatile RAM and ROM memory, running a mufti-tasking
operating
system. Typically FLASH memory for program store and application store. The
units will
typically have a plurality of serial and parallel, electrical and optical
interfaces for
machine-to-machine and machine-to- person communications. The unit is equipped
with
dedicated control electronics for controlling and interfacing to the waveslot
alignment
units. The unit can be optionally fitted with a shared optical monitor system
with optical
power, wavelength, OSNR, Q monitor etc capability. The application store
contains
application software for control, maintenance, status monitoring, performance
monitoring,
and network management protocols. Communication protocols, alarm detection and
reporting, protection and restoration, communications and control loops for
management
the waveslot alignment, power equalization, optical power fill transmitters
etc. The switch
node may be controlled and monitored by a software running on a remote
computer, say
using a telnet session over TCP/IP, or by a network operating system
Figure 15 is similar to figure 14 in operation except that the inlet and
outlet fibers have
been interleaved to fold the plane switch fabric 278 to permit hair-pinning
and loop-back of
waveslots using a single 4x4 switch.
The switch unit 278 of figure 15 is shown in more detail in figure 16. In this
example the
eight 8x8 optical space switch modules 294 form a wavelength plane that is
fully folded by
having both waveslots and/or wavelengths from inlet and outlet fibers
interleaved passing
through the space switches permitting hair-pinning and loop-back of waveslots
on the same
switch device. During operation the switch operates in a plurality of states
based on
combinations of the BAR state, the CROSS state and the ISOLATE state the basic
building
blocks of the switch. As larger switches become commercially available they
can be
incorporated in this architecture.
In figure 17, illustrates a method for connecting an aggregation node a switch
node via the
facility (tandem) fibers- facility fibers that typically connect to other
switch nodes in a
mesh network. In the example aggregation nodes 312, 310 connect to switch node
304 on
316, one of the inlet fibers 317 to the switch node and to one of the outlet
fibers 306 of 305
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
29
from the switch node. In this example node 310 output add port 31 l and node
312 output
add port 3 L3 connects to 315 that connects to 3I6. 315 can be just a splice
or a combiner or
it can be incorporated into the aggregation node. The switch node outlet fiber
306 connects
to 307, which connects to node 312 input drop port 308a and node 310 input
drop port
309a. 307 can be just a splice or a combiner or it can be incorporated into
the aggregation
node. Port 308b on node 312 and port 309b on node 310 can be connected in a
similar
manner to switch node 304 on another set of inlet and outlet fibers for
optical link
redundancy, or ports 308b and 309b can be connected to another switch node for
matched
switch node operation for site redundancy, or ports 308b and 309b may be left
unconnected, and used for test access etc.
In figure 18, illustrates a method for connecting an aggregation node to the
switch units,
278 in figure 14 and 15, and 294 in figure 16, of a co-located switch node,
bypassing the
DEMUR and MUX units, therefore avoiding the use of valuable facility (tandem)
fibers
that instead can be used to connect to other switch nodes or more remote
aggregation
nodes. In the example aggregation node 322 output and input port 323, both
primary 323a
and secondary ports 323b for full link redundancy, connect to the switch units
of the switch
node via links 324 that connect to the links 279 of the switch units.
Aggregation node 321
primary output and input port 320a connect to the switch units of the switch
node via links
3l 9 that connect to the links 279 of the switch units. The secondary port
320b of node 321
can be connected in a similar manner to switch node 304 to 318 for optical
link
redundancy, or the secondary ports 320b can be connected to another switch
node for
matched switch node operation for site redundancy, or secondary ports 320b and
switch
port 318 may be left unconnected, and used for test access etc.
Figure 19 shows how the DEMUR and MUX units of a 4-fiber aggregation node, (as
show
in figure I 3) are typically connected. Pass-through port 271 of DEMUR unit
272 is
connected passively with fiber 325b or actively with a compensating unit 325a
to inlet
pass-through port 276 of MUX unit 275. Pass-through port 251 of DEMUR unit 250
is
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
connected passively with fiber 326b or actively with a compensating unit 326a
to inlet
pass-through port 256 of MUX unit 236.
In figure 20, shows how the DEMUR and MUX units of a 4-fiber aggregation node
that
are collocated can be connected to minimize pass-through loss. For the inlet
fiber 331 to
outlet fiber 328 the WEST DEMUR units 250 are serially connected from the
aggregation
nodes than the EAST MUX units 236. Likewise for the opposite direction for
inlet fiber
337 to outlet fibre 332 the EAST DEMUR units 272 are serially connected from
the
aggregation nodes than the WEST MUX units 275.
The system management to ensure the phase of the waveslots generated by the
transmitter
is aligned to the optical switch node timing, when the waveslots arrive at the
optical switch
controls the precise timing of transmitter output. Waveslots arriving from
other switch
nodes are phased appropriately by propagation through switched fibre delay
line systems,
which align the waveslots to the switch operation, (units 290 in figures 14
and 15) using
the arrangement illustrated in Figure 21. Until optical delay devices become
commercially
available a straightforward approach is to form the desired delay using fixed
fibre lengths,
arranged in an exponential sequence, that are switched in and out to introduce
the desire
delay for the duration on the fibre span.
Typical packaging for such delay elements is shown in Figure 21. 351 is an
array of 10
fibre based delay elements 349. Each delay element 353 is a loop of fibre. The
switches for
switching in and out the loops are contained in 351.The target is less than 2
dB insertion
loss per element.352 is the input, 350 is the output. An alternate packaging
geometry is
shod with 361. A more compact packing is shown with 357 where 354 is fibre
delay loop,
356 '. the input and 355 is the output. The switches are packaged inside 357.
The
packa~'ging for the alignment unit is similar to that of commercially
available fiber based
chromatic dispersion compensation modules.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
31
The delay can be both locally optimized, however this approach uses end-to-end
optimization. Which is more complete and exact in terms of being able to
adjust for
variations in network topologies that affect the amount of delay that needs to
be introduced.
It also permits consideration to be given for reducing the specifications,
absolute or
relative, or the specification tolerance o other network components and
elements that affect
delay.
The Line of Sight control (LOSP) Protocol that controls the setup, management
and
takedown of a 'connection' is shown by a simplified example in figures 22 to
28.
)figure 22 shows how the LOSP is designed to try the shortest and least
congested path
first to send data traffic from the source node 381 to the destination node
366. Assigning
two specific link meRics to each link does this. The first is a percentage
utilization
(example: UMSSZ in figure 22) that is updated every 5 or 10 minutes or so from
information
sent by the switch nodes 383,362,364,370,379 and 368 to all the aggregation
nodes in the
line of sight. The second is a nominal distance (ea;ampte: n".,ssZ in figure
22), which is
provisioned at start up to reflect link cost based on distance between nodes.
The original
copy of this link state information is kept locally at the switch node. The
link state
information can be sent to a node in the event of it re-joining the network
after restart or
upon initial first connection.
Figure 23 shows an example of a connection request. The end user on the device
396
requests a connection for NxOC-3s from the 396 connected to aggregation node
381
(labeled Ms) over fibre 397, to the multiplexer 385, connected to destination
aggregation
node 366 (labeled MD) over fibre 384. Source node 38i computes Dijkstra to
determine
shortest nominal path to 366. Dijkstra link cost parameter is a product of
basically the
percentage of link utilization and nominal distance. The example route 394 to
391 to 387,
over fibre 380, 378 and 367 respectively through nodes 379 (S2) and 368 (SS)
is identified.
At this point the protocol does not know if any channels over waveslots are
available over
this path. The Line of sight protocol finds open channels from source to
destination over
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
32
this pith. If an open channels) is found, the management process at node 381
sends a
connection seeking message/packet 395 to the first switch node 379 on the
selected path
(route). The message/packet can be sent in-band via a management waveslot to
the switch
node or out of band via a separate IP control network. The nodal processor at
switch node
379 updates the message/packet and sends it 390 on to the next switch node
368, which
updates it as well and sends it 386 where it reaches the destination node 366.
Figure 24 shows the format of the LOSP connection-seeking packet (CSP). 398 is
the
connection seeking packet identifier (CSP1), 399 is the blocking link (BL) and
400 is the
line of sight state (LSS). The CSP is encapsulated in an IP packet and
transmitted from
node to node along selected route. The CSP is updated at each node before it
is
retransmitted to the next adjacent node. The CSPI 398 is 56 bytes long and
contains the
source node identification (Ms), the destination node identification (M~),
selected route,
VPNID, Priority, Bandwidth (BW), Time of request, and Blocking event register.
The BL
399 is 4 bytes long. The BL identifies the "Most Blocking Link" encountered so
far. The
BL is used to eliminate the worst link in the event of blocking. The LSS 400
is 320 byes, ar
2560 bits, one bit for each of 64 timeslots and colour combination. Bit =
logic 0 indicates
open timeslot and colour (waveslot) position. The CSP is originated at the
source node 381
with the LSS having a full set of connection possibilities available and is
updated at each
intermediate switch node on the way to the destination node 366 by being
logically OR'ed
with the Link Sate at that node.
Figure 25 illustrates how the LOSP identifies the open channels on the best
route between
source and destination points. The Line of sight link state matrix is
progressively occluded
as the CSP traverses the route from the source node 381 to the destination
node 366. At
each node the number of channels blocked is calculated and the blocking link
field, BL,
399 in figure 24 is calculated and the field is updated if the new link is
worst than other
previous traversed links on the route.
CA 02440004 2003-09-05
WO 02/071791 PCT/CA02/00301
33
Figure 26 illustrates how the channels are reserved for a connection. The
destination node
366 randomly selects as many channels as indicated in the bandwidth (BV~ field
of the
CSPI, 398 in Figure 24. Node 366 then encapsulates the information into a
Reservation
request packet (RRP) 386 and transmits to the first switch node 368 on the
reverse route.
The switch node 368 reserves the channels for the connection. The first switch
node 368
transmits an RRP 390 to the second switch node 379 on the route, and it
likewise reserves
channels for the connection. The source node 381 for the connection receives
an RPP
packet 395 from node 379 and sends an acknowledgement (ACK) packet 409 to the
destination node 366. The destination node 366 in turn sends an
acknowledgement (ACK)
packet 410 to the destination node 381 and transmission on the connection path
over fibers
380, 378 and 367 begins.
When the RRP arrives at a switch node, and the nodal processor at the node
finds that the
requested channels have been already taken, meaning a "colliding " RRP got
there, the
nodal processor updates the "Line of Sight State" and returns the RRP to the
destination
node 366. The destination node then randomly selects new channels and launches
a new
RRP. The channels are reserved if possible, if not the destination node
selects new channels
and a new RRP is launched until an available path is found and a RRP arrives
at the source
node 381, and the channels are reserved along the path and the connection
session can
begin.
Figure 27 shows the format of the LOSP reservation request packet (RRP). 411
is the
connection seeking packet identifier (CSPI), 412 is the blocking link (BL),
413 is list of
selected channels (LSC) and 414 is the line of sight state (LSS).
Figure 28 is ifthe route is blocked because ofa failure 420, the source node
381 eliminates
the offending links) 378 from the node's topology map using the "blocking link
field" 412
of figure 27. The management system re-calculates the Dijkstra and then re-
initiates the
process with the next best route, in this case 421,415,416,418 and 419,
through switch
nodes 383,362,370 and 368 to the destination node 366.
