Note: Descriptions are shown in the official language in which they were submitted.
CA 02276518 1999-06-30
WO 98131159 PCTII1S98/00058
MESH NETWORK WITH HIGH RESTORATIVE CAPACITY
Background of the Invention
This invention relates to telecommunications
networks. More particularly, this invention relates to
apparatus and methods for improving the reliability of
mesh telecommunications networks. Improved reliability
is achieved by providing high restorative capacity for
restoring network integrity should portions of the
network become inoperative.
Networks are ubiquitous; they are the
i5 backbone of many services and conveniences. For
example, automated teller machines are part of banking
networks that conveniently increase access to banking
services. Many modern retail cash registers are part
of a network used by retailers to track sales, set
prices, and maintain inventory. Telephone, computer,
and cable TV systems are all further examples of
services made possible by telecommunication networks.
Common to these networks is the transport of
some kind of information, in one form or another, from
a source to a destination. This information, which can
represent, for example, computer data, voice
transmissions, or video signals, is known as "traffic."
Traffic enters a network usually at a node, is
transported through the network via links and other
nodes until a destination is reached, and then exits
the network usually at another node. Nodes provide the
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCT/US98/00058
- 2 -
routing necessary to either input (or "add") new
traffic to the network, output (or "drop") traffic from
the network, or direct traffic from one portion of the
network to another. Links provide traffic paths
between nodes. Overseeing the operation of a network
is some kind of control. Control may be centralized,
where all traffic management decisions are made by a
central controller, or decentralized, where individual
nodes have limited traffic management capabilities.
Networks vary in size and complexity. For
example, a network could consist of a handful of
computers connected together in a single office, or
could consist of millions of telephone customers
connected together across a continent.
Network configurations also vary. For
example, a "mesh" network is one in which most nodes
are connected to three or more other nodes. A
symmetrical mesh network, as shown in FIG. lA, results
when each node is connected to an equal number of other
2o nodes (except at the periphery of the network). An
asymmetrical mesh network, as shown in FIG. 1B, results
when nodes are connected to a variable number of other
nodes. A ring network, as shown in FIG. 1C, is an
interconnection of "rings," in which nodes and links
are connected in a circular fashion.
Links can be of various transmission media,
but more commonly, are either fiber-optic cable or
coaxial cable. Individual links can vary in length
from a few feet to hundreds of miles. Links that are
3o part of a larger network, such as a telephone system,
are usually carried on overhead utility poles, in
underground conduits, or in combinations of both.
Nodes can range in complexity from simple
switching or relay devices, as may be found in smaller
networks, to entire buildings containing thousands of
devices and controls, as may be found in larger
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCT/US98/00058
- 3 -
networks. Nodes can be implemented electronically,
mechanically, optically, or in any combination thereof.
' Nodes, generally known as cross-connects,
perform a variety of functions. They perform basic
~ 5 traffic routing such as adding traffic to, dropping
traffic from, and directing traffic through the
network. Nodes also provide status to a network
control system. In those networks where control is
centralized, nodes simply transmit status to, and
id execute instructions from, the control system. In
those networks with decentralized control, nodes are
more complex enabling them to communicate with other
nodes and make traffic routing decisions. Thus, nodes
serve a variety of purposes based on the type of
15 network control and the particular needs of a given
network location.
The amount of data transported by a network
can be very large. Typical data transfer rates for a
fiber-optic link can range from 2.5 gigabits per second
2o to 10 gigabits per second. A "bit" is a binary digit,
which is the basic unit of computer data. A "gigabit"
is a billion bits. Accordingly, any disruption in
network traffic flow can be devastating. Of particular
concern are telephone networks, where hundreds of
25 thousands of individual communications could be
transporting through the network simultaneously. Thus,
network reliability, that is, the continuous
availability and operation of a network, is commonly a
top priority of network operators.
3o Network control and link integrity are two
areas that can have the greatest impact on network
reliability. For example, a control system malfunction
is likely to affect some, if not all, of a network's
performance. Link failures cause tremendous traffic
35 losses (2.5 to 10 gigabits per second). Thus, to
improve network reliability, backup control systems
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCTILlS98/00058
- 4 -
must be provided to maintain control should the main
control system fail, and spare links should be
installed to permit rerouting of traffic disrupted by
link failures.
One known mesh network that includes such
reliability features is a long distance telephone
network. A central controller monitors and controls
the entire network, and several back-up systems ensure
continuous operation. Each node communicates with the
to controller, sending status and receiving instructions
for properly routing traffic. Working links connect
the nodes and provide dedicated pathways for
transporting traffic. A number of spare links, which
do not regularly transport traffic, are installed in
l5 particular areas to provide alternative pathways for
rerouting traffic that has become disrupted by an
inoperative working link.
A link can become inoperative in a number of
ways, but most often, when it is cut. This usually
20 occurs, for example, when excavation occurs over an
underground link, or when a traffic accident or severe
storm damages a utility pole carrying a link.
Although these incidents are rare, typically
occurring only about once per year, when one does
25 occur, the nodes connected to the inoperative link
immediately notify the controller. The controller then
determines whether either enough spare links, spare
capacity on working links, or combinations of the two,
are available to reroute the disrupted traffic. Once
3o an alternative traffic path is determined, the
controller then sends appropriate instructions to those
nodes that can interconnect the identified spare links
and working links to form the alternative traffic path.
Typical recovery time from such a disruption is
35 approximately two seconds. This recovery time was once
hailed as a marvel of technology; today, however, it is
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 pC~i.~s,~g~g
- 5 -
no longer acceptable. A two-second outage would
adversely affect, for example, the transmission of
' computer data. In fact, an entire computer center
could be adversely affected by such an outage.
Contributing to the unacceptable recovery
time is the network's centralized control. Nodes in
such a network do not usually have the capability of
communicating directly with other nodes to restore
traffic disruptions. Thus, nodes affected by a link
disruption must first communicate the disruption to the
central controller, await instructions, and then, along
with selected adjacent nodes, execute the received
instructions to form the alternative path. This
process undesirably increases recovery time.
~5 Further, because centralized control renders
a network more susceptible to network-wide disruptions
should a malfunction occur, complex and expensive
network-wide backup systems are needed to protect
against such possibilities. This undesirably increases
the cost of equipment, personnel, and maintenance for
this type of network.
Another known network that improves upon the
mesh network described above is a ring network. Nodes
are connected in a circular fashion to form rings, and
multiple rings are interconnected to. form the complete
network. Nodes are either add/drop multiplexers (ADMs)
or cross-connect switches. An ADM adds or drops
traffic from the network or simply forwards traffic to
the next node. A cross-connect switch interconnects
one ring with another. Control in this network is
decentralized, enabling nodes to make limited traffic
routing decisions. Although the rings are
interconnected, each ring operates independently of the
others thus desirably reducing the possibility of a
network-wide failure.
One ring of such a network is shown in
SUBSTITUTE SHEET (RULE 2B)
CA 02276518 1999-06-30
WO 98/31159 PCT/US98100055
- 6 -
FIG. 2. Ring 200 includes nodes 202, 204, 206, 208,
210, and 212. The connections between each node are
made with one working link and one spare link. If a
working link becomes inoperative (e.g., is cut), the
traffic transported by that link will be rerouted back
around the ring through the spare links.
To illustrate, assume a break 214 occurs
between nodes 202 and 204, as shown in FIG. 2, severing
working link 215 and spare link 216. Nodes 202
and 209, sensing the disruption of traffic flow,
communicate with adjacent nodes 212 and 206,
respectively, which in turn communicate with adjacent
nodes 210 and 208, respectively. These nodes then
activate spare links 218, 220, 222, 229, and 226 to
t5 reroute traffic to nodes 202 and 209. Recovery time
from such a disruption is typically in the microsecond
to nanosecond range. A microsecond is a millionth of a
second and a nanosecond is a billionth of a second. In
that short amount of time, telephone customers would
not realize that a link carrying their call was cut and
rerouted, and transmitted computer data would likely
suffer only the loss of a few bits of data, which would
simply require retransmission of the lost bits. Thus,
this network improves upon the performance of the
previously described mesh network.
However, a disadvantage of this ring network
is that restoration is limited to substantially only
one inoperative working link per ring. If, for
example, two working links were cut in the same ring,
traffic flow could not be restored until at least one
of the links was physically repaired. (One exception
is the case where one of the two inoperative working
links occurs between the nodes of an interconnecting
ring, as shown in FIG. 3. Traffic can be restored by
including spare link 324 of ring 322 with spare
links 306, 308, 310, 312, 314, and 316 of ring 302 to
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCT/US98/00058
- -7 _
restore traffic disrupted by breaks 301 and 303.
However, such an occurrence would be completely
fortuitous and does not significantly diminish the
severity of this disadvantage.) This disadvantage is
' S not shared by the previously described mesh network,
because most nodes in a mesh network are connected to
three or more other nodes, increasing the likelihood
that enough spare links or spare capacity would be
available to completely restore traffic flow that was
disrupted by multiple inoperative working links.
A further disadvantage of this ring network
is the high percentage of links that are set aside as
spare -- a full SOo. Thus, half the links in the
network will either sit idle, or, at best, be
~5 underutilized with nonessential or low priority
activity until needed to restore disrupted traffic
flow. This high percentage of underutilized link
capacity is undesirable in today's environment of ever
increasing demand for computing and communications
power and flexibility, which accordingly increases
demands on network resources arid reliability.
Consequently, a mesh network, with its
greater number of nodal interconnections, appears to
provide a better framework from which to improve
network reliability without unduly burdening the
network with a high percentage of underutilized
resources.
In view of the foregoing, it would be
desirable to provide a mesh telecommunications network
with high restorative capacity for restoring network
. traffic flow should one or more working links become
inoperative.
. It would also be desirable to provide a mesh
telecommunications network with a sufficient number of
spare links for restoring disrupted traffic flow while
reducing underutilized network resources.
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCT/US98/OOOS8
_ g _
It would further be desirable to provide a
mesh telecommunications network with decentralized
control to improve restoration time and to reduce the
likelihood of network-wide failures.
Summary of the Invention
In accordance with this invention, there is
provided a mesh telecommunications network with
i0 improved reliability for transporting traffic from a
source to a destination. The network has high
restorative capacity and includes nodes for adding,
dropping, and directing traffic, and links for
transporting traffic between nodes. Each connected
pair of nodes has at least three links between them; at
least two, known as working Links, provide dedicated
traffic transport, and at least one, known as a spare
link, provides selectable alternative traffic transport
should a working link become inoperative.
The network advantageously has decentralized
control for reducing the likelihood of network-wide
failures and for improving restoration time.
Decentralized control improves restoration time by
enabling nodes affected by inoperative working links to
communicate directly with adjacent nodes to quickly
establish alternative paths comprised of spare links.
The allocation of spare links throughout the network is
generally sufficient to provide complete restoration of
typically disrupted traffic flow, while also reducing
the typical amount of underutilized link capacity.
Brief Description of the Drawings
The above and other objects and advantages of
the invention will be apparent upon consideration of
the following detailed description, taken in
conjunction with the accompanying drawings, in which
like reference characters refer to like parts
SUBSTITUTE SHEET (RULE 26j
CA 02276518 1999-06-30
WO 98/31159 PCTIUS98I00058
_ g _
throughout, and in which:
FIGS. IA, 1B, and 1C are each a
representational diagram of a network configuration;
FIG. 2 is a representational diagram of a
' S portion of a prior art ring network;
FIG. 3 is a representational diagram of a
larger portion of a prior art ring network;
FIG. 9 is a representational diagram of a
link connecting two nodes;
FIG. 5 is a representational diagram of a
portion of a first embodiment of the present invention;
FIGS. 6A and 6B are representational diagrams
of a portion of a preferred embodiment of the present
invention:
15 FIG. 7 is a representational diagram of a
portion of a third embodiment of the present invention;
and
FIG. 8 is a representational diagram of a
portion of a fourth embodiment of the present
20 invention.
Detailed Description of the Invention
The present invention provides a mesh
25 telecommunications network with improved reliability.
The network transports information, known as "traffic,"
in one form or another, from a source to a destination.
Traffic can represent, for example, computer data,
voice transmissions, or video signals. The network
30 includes a plurality of nodes and links. Nodes route
traffic into the network, out of the network, and from
one portion of the network to another. Such nodes are
generally known as cross-connects. Links interconnect
the nodes to provide a system of traffic paths, each
35 link being connected to two nodes. A "mesh" network is
configured such that most nodes are connected via links
to three or more other nodes. Examples of mesh
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCTIUS98100058
- 10 -
networks are shown in FIGS. lA and 1B.
Traffic enters the network usually at a node,
is transported via a plurality of links and other nodes
to a destination, and then exits the network usually at
another node. "Traffic," as used herein, refers both
to a single communication being transported through the
network from a source to a destination, and to all
communications being transported through the network
from a plurality of sources to a plurality of
destinations.
Links are advantageously fiber-optic cable
for transport of traffic in optical signal form. Other
transmission media, such as, for example, coaxial cable
for electronic signal transport, could also be used.
Each link provides two separate paths for transporting
traffic between two nodes. As shown representationally
in FIG. 4, link 401 has a first path 402 for
transporting traffic from a first node 404 to a second
node 906, and a second path 408 for transporting
2o traffic from second node 406 to first node 404. For
simplicity, each link is shown in the drawings as a
double-headed arrow.
Links are allocated as working links and
spare links. Most links are working links that provide
dedicated traffic transport between the two
nodes connected thereto. Spare links, which do not
normally transport traffic, provide selectable
alternative traffic transport for restoring traffic
flow between nodes that have had one or more working
links between them become inoperative. Thus, spare
links, while enhancing network reliability, also
constitute an underutilized network resource.
Therefore, providing a sufficient number of spare links
such that the network is adequately protected and yet
not unduly burdened is one of the more advantageous
features of the invention.
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCTIUS98100058
- 11 -
Nodes are complex structures containing
thousands of devices and controls for routing and
preferably managing traffic flow. The design of such
nodes, and the components used within, are well known
in the art. Nodes are implemented preferably
electronically, but can also be implemented, for
example, optically, mechanically, or in any combination
thereof. Besides performing basic traffic routing
functions, nodes are also in communication with a
network controller via links, providing status and
other control information.
Network control is advantageously
decentralized, enabling nodes to communicate with
adjacent nodes and make limited traffic routing
i5 decisions. This reduces the time needed to restore
traffic flow disrupted by inoperative working links,
because the nodes affected by the disruption can
cooperate directly with adjacent nodes to establish
alternative traffic paths, rather than having to first
communicate the disruption to a controller, await
instructions while the controller, which is likely
handling other tasks as well, determines an alternative
path, and then execute the received instructions.
Decentralized control also reduces the
likelihood of network-wide failures. By distributing
traffic management functions to nodes throughout the
network, problems arising in a controller, such as
hardware failures or software errors, are much less
likely to affect the entire network.
Restoration of disrupted traffic flow is
advantageously accomplished by connecting together a
minimum number of spare links to form one or more
alternative traffic paths to the nodes affected by
inoperative working links. Generally, such
inoperability occurs when a link has been cut or
severed, such as when excavation cuts through an
SU9STtTUTE SHEET (RULE 28)
CA 02276518 1999-06-30
WO 98/31159 PCTIUS98/00058
- 12 -
underground conduit carrying a link, or when severe
weather severs a link being carried on an overhead
utility pole. An inoperative working link is sensed by
the two nodes connected to that link. The two affected
nodes then communicate the disruption to the network
controller so repairs can be scheduled, and then cause
the traffic from the inoperative links to be routed to
spare links. Communication from the affected nodes to
adjacent nodes is accomplished via spare links. The
1o communication is detected by receivers in the spare
links that cause the adjacent nodes to activate the
appropriate switches to connect the spare links with
other spare links to form the alternative traffic path.
Programming within the nodes selects the most direct
available alternative path.
Typical recovery times from such link
disruptions are desirably in the microsecond to
nanosecond range, dependent, in part, on the switching
technology of the nodes.
A portion of a first embodiment of a network
according to the present invention is shown in FIG. 5.
Network 500 has a plurality of nodes that are
advantageously electronic, and a plurality of
links that are advantageously fiber-optic, connected in
a symmetrical mesh configuration. Symmetry results
from each node being connected to an equal number of
other nodes (except at the periphery). Control of
network 500 is advantageously decentralized. Each node
can sense the operability of the links connected to it,
and can communicate with adjacent nodes and the
controller. Each connected pair of nodes has three
links therebetween. Two of the Iinks are working
links and the other is a spare link. Thus, there is a
spare link between every pair of connected nodes and
only one-third of all links are spare links. This
allocation of spare links is a 33.3 improvement in
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 pC~L/USgg~000sg
- 13 -
underutilized link capacity as compared to the
previously known ring network.
When network traffic is disrupted because of
one or more inoperative links, traffic is restored as
follows: assume a break 501 occurs between nodes 502
and 504, as illustrated in FIG. 5, severing working
links 505 and 507 and spare link 506. Restoration of
severed spare link 506 is unnecessary because it does
not regularly transport traffic; thus there is no
traffic flow to restore. To restore traffic flow from
severed working links 505 and 507, nodes 502 and 504,
acting in concert, first communicate the break via
links to the controller (not shown) and then begin to
substantially simultaneously form alternative traffic
IS paths 510 and 520. Alternative traffic path 510 is
made up of spare links 512, 515, and 517 and node
switches 514 and 519. Alternative traffic path 520 is
made up of spare links 522, 525, and 527 and node
switches 524 and 529.
2Q Alternative traffic path 510 is formed by
node 502 communicating with node 513 via spare link 512
to activate switch 519. Switch 514 connects spare
link 512 with spare link 515. Meanwhile, node 504
communicates with node 518 via spare link 517 to
25 activate switch 519. Switch 519 connects spare
link 517 with spare Link 515, thus completing
alternative traffic path 510.
Alternative traffic path 520 is formed
similarly. Node 523, after receiving communication
3o from node 502 via spare link 522, activates switch 524
to connect spare link 522 with spare link 525.
Meanwhile, node 528, after receiving communication from
node 504 via spare link 527, activates switch 529 to
connect spare link 527 with spare link 525, thus
35 completing alternative traffic path 520. Traffic flow
previously provided by severed working links 505
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCTIUS98100058
- Z4 -
and 507 is now restored to nodes 502 and 504.
The present invention is particularly
effective in networks having greater numbers of links
between nodes. For example, a portion of a preferred
embodiment of a network according to the present
invention is illustrated in FIGS. 6A and 6B.
Network 600 is a symmetrical mesh network with
decentralized control of traffic flow. In this
network, each connected pair of nodes has four links
connected therebetween, three working links and one
spare link. Accordingly, every pair of connected nodes
has a spare link connected therebetween, and only one-
fourth of all links are spare links. This allocation
of spare links represents a 50o improvement in
underutilized link capacity as compared to the
previously described ring network.
To facilitate explanation and understanding
of the restoration process according to the principles
of the present invention, each spare link in FIGS. 6A
and 6B is shown as two separate unidirectional paths,
each represented by a single-headed arrow, which
indicates the direction of traffic flow.
In this preferred embodiment, spare links are
preferably pre-connected in a standby mode as shown in
2S FIG. 6A. For example, spare link 602a is connected by
switches 623 and 603 to spare links 692b and 612a,
respectively. Spare links 642b and 612a are then
connected to spare link 632b via switches 643 and 633,
respectively, to form a selectable unidirectional
standby alternative path between nodes 610, 630, 640,
and 620. Such selectable standby alternative paths are
formed between each group of nodes. These standby
paths significantly improve restoration time by
providing established alternative traffic paths for
substantially immediate transport of disrupted traffic
flow. Furthermore, these standby paths can be modified
SUBSTITUTE SHEET (RULE 26j
CA 02276518 1999-06-30
WO 98/31159 PCTlUS98100058
- 15 -
as needed by reconnecting the node switches to other
spare links to form other alternative paths.
To illustrate the restoration of disrupted
traffic flow in network 600, assume a break 601 occurs
between nodes 610 and 620, disrupting traffic flow in
working links 611, 613, and 615, as shown in FIG. 6B.
To restore traffic flow, six unidirectional alternative
traffic paths must be formed, three transporting
traffic from node 610 to node 620, and three
t0 transporting traffic from node 620 to node 610.
Two of the six alternative paths are
substantially immediately available via standby
alternative paths; one standby path includes spare
links 612a, 632b, and 642b, and the second standby path
includes spare links 622b, 662a, and 652a. Thus
traffic transported by one of the disrupted working
links 611, 613, or 615 is substantially immediately
restored by the two unidirectional standby paths.
Programming within the nodes determines what traffic
from the severed working links is transported by the
standby paths.
A third alternative traffic path is
preferably formed as follows: node 640, after receiving
communication from node 620 via spare link 642a,
activates switch 645 to connect spare link 642a with
spare link 632a. Substantially simultaneously,
nade 630, after receiving communication from node 620
via spare link 612a, activates switch 635 to connect
spare link 632a with spare link 612b, thus completing a
third alternative traffic path between nodes 620
and 610.
Similarly and substantially simultaneously as
the third alternative path, a fourth alternative
traffic path is preferably formed as follows: node 650,
after receiving communication from node 610 via spare
link 652b, activates switch 655 to connect spare
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCT/US98100058
- 16 -
link 652b with spare link 662b. Meanwhile, node 660,
after receiving communication from node 620 via spare
link 622b, activates switch 665 to connect spare
link 662b with spare link 622a, thus completing a
fourth alternative traffic path between node 610
and 620.
A fifth alternative traffic path is formed
also substantially simultaneously as the third and
fourth alternative paths preferably as follows:
node 630, after receiving communication from node 610
via spare link 612a, activates switch 637 to connect
the standby alternative path formed by spare
links 614a, 672a, and 634b with the standby alternative
path formed by spare links 682a, 684b, and 686b.
Meanwhile, node 640, after receiving communication from
node 620 via spare link 642a, activates switch 647 to
connect the standby alternative path formed by spare
links 682a, 684b, and 686b with the standby alternative
path formed by spare links 644b, 692b, and 624a, thus
2o completing a fifth alternative traffic path.
A sixth alternative path is formed
substantially simultaneously as the other alternative
paths preferably as follows: node 660, after receiving
communication from node 620 via spare link 622b,
activates switch 667 to connect the standby alternative
path formed by spare links 624b, 694b, and 664a with
the standby alternative path formed by spare
links 666b, 696a, and 654a. Meanwhile, node 650, after
receiving communication from node 610 via spare
link 652b, activates switch 657 to connect the standby
alternative path formed by spare links 666b, 696a, and
659a with the standby alternative path formed by spare
links 656a, 674x, and 614b, thus completing a sixth
alternative traffic path.
Complete traffic flow, previously transported
by severed working links 611, 613, and 615, is now
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCT/US98I00058
- 17 -
restored to nodes 610 and 620. Restoration time is
significantly improved over the prior art mesh network
because of decentralized network control, sufficient
numbers of spare links, pre-connected standby
. 5 alternative paths, the accordingly limited number of
neighboring nodes that needed to be communicated
with -- in this case only four, nodes 630, 690, 650,
and 660, and the limited amount of switching those
neighboring nodes needed to perform to complete the
alternative paths.
Furthermore, if any of the spare links that
were used to form the alternative paths above had been
unavailable (except the spare links connected directly
to nodes 610 and 620, which must be available for
is complete restoration), it is likely that other spare
links could have been used to form the alternative
paths. The use of other spare links would likely
result in longer alternative paths and longer
restoration times because of the additional nodes to be
communicated with and switches to be set, but
alternative paths nonetheless. This improved
restorative capacity clearly demonstrates one of the
advantages of a mesh network over a ring network, where
restoration is limited to typically one alternative
path per ring.
The present invention is also effective in
asymmetrical mesh networks, which have nodes that are
not each connected to an equal number of other nodes.
FIG. 7 illustrates a portion of a third embodiment of a
mesh network according to the present invention.
Network 700 is an asymmetrical network with
decentralized control and a plurality of nodes,
preferably implemented electronically, interconnected
with a plurality of links, which are advantageously
fiber-optic cable. Each connected pair of nodes has
three links therebetween, two working links and one
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCTIUS98J00058
- 18 -
spare link. Spare links are again shown as bi-
directional paths represented by double-headed arrows.
Note that the feature of pre-connected standby
alternative paths could also be included here, and that
the same principles of the invention would apply if
there were four links between each connected pair of
nodes, three working links and one spare link.
Referring to FIG. 7, restoration of traffic
flow is as follows: assume a break 701 severs the links
r0 between nodes 710 and 720. Alternative traffic
paths 731 and 753 can be formed substantially
simultaneously to restore the traffic flow of working
links 711 and 713 in the same manner as previously
described for the embodiments shown in FIGS. 5 and 6B.
is Alternative path 731 is formed by connecting spare
link 732 to spare link 734 via switch 736 at node 730.
Spare link 739 is connected to spare link 742 via
switch 744 at node 740, thus completing alternative
path 731. Similarly, alternative traffic path 753 is
20 formed by connecting spare link 752 to spare link 756
via switch 754 at node 750. Spare link 756 is
connected to spare link 766 via switch 762 at node 760.
Spare link 766 is connected to spare link 779 via
switch 772 at node 770. Spare link 774 is connected to
25 spare link 784 via switch 782 at node 780, thus
completing alternative path 731.
As can be seen, even from the limited portion
of network 700 shown, variations of alternative
paths 731 and 753 are possible by connecting other
30 spare links through other nodes. If, for example,
spare link 766 had not been available, alternative
path 753 could have been routed through node 740 with
spare links 764 and 746. Moreover, depending on the
interconnection of nodes not shown in FIG. 7, other
35 alternative paths for restoring traffic flow could have
been possible. Generally, the more nodal
SUBSTITUTE SHEET (RULE 2fi)
CA 02276518 1999-06-30
WO 98!31159 pCTIUS~~00p58
- 19 -
interconnections there are, the more likely it is that
several alternative paths are available, increasing the
likelihood of complete restoration should multiple link
failures occur.
S FIG. 8 illustrates a portion of a fourth
embodiment of a mesh network according to the present
invention. Network 800 includes a plurality of
electronically implemented nodes 802 interconnected by
a plurality of fiber-optic links 809. However, instead
of each link 804 transporting traffic sequentially, as
in the previous embodiments, each link 804 transports a
plurality of traffic in parallel. Parallel traffic
transport is accomplished by transporting each
plurality of traffic through the link at a unique
1S transporting parameter. This parameter is preferably
wavelength and the manner in which transport is
accomplished is wavelength-division-multiplexing (WDM),
which is known in the art.
Typically, up to eight wavelengths per link
20 are possible, four wavelengths for each direction
(i.e., four for transport from a first node to a second
node, and four for transport from the second node to
the first node). Note that while the number of
wavelengths per link may likely increase with
2S advancements in the state of the art, the principles of
the present invention would still apply.
Each pair of connected nodes is connected by
a single link, which is capable of transporting traffic
at approximately 20 gigabits per second. Wavelength
3o multiplexers 806, located at each node, provide the
necessary wavelength modulated traffic multiplexing and
demultiplexing, and translation from optical signal
form to electronic signal form and vice versa.
Restorative capacity is established by
3S setting aside at least one wavelength per direction as
a spare. Thus, traffic can be transported at three
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98131159 PCT/US98100058
- 20 -
wavelengths per direction per link. Accordingly, the
percentage of "spares" can be as low as 25% of the
total number of wavelengths available for transporting
traffic, the same percentage of spare links as in the
preferred embodiment of the present invention. Spare
wavelengths are then available for use in selectable
alternative traffic paths. Nodes still provide the
connections between links for transporting disrupted
traffic, and the wavelength multiplexers provide the
proper routing of wavelength modulated traffic into and
out of the nodes. Furthermore, standby alternative
paths can also be provided by appropriately presetting
the multiplexers and node switches to accommodate
traffic flow at a spare wavelength. Such paths would
enable disrupted traffic flow to be substantially
immediately rerouted. Forming additional or modified
alternative traffic paths is then accomplished in the
same manner as previously described for the embodiments
shown in FIGS. 5, 6B and 7.
In each of the above embodiments, it is
possible that several inoperative links may limit a
network's ability to completely restore disrupted
traffic flow. For example, referring to FIG. 6B, if a
second break were to occur between nodes 610 and 630,
complete restoration of traffic flow to and from
node 610 could not be made. In these rare situations
(recall that only one such break typically occurs per
year in the known mesh network described previously),
those traffic paths with a pre-determined higher
priority will be restored. Network areas deemed
critical or more vulnerable can be supplemented with
either more nodal connections or more spare links to
enhance the likelihood of a complete recovery under
such atypical circumstances. Note that while
additional spare links between connected pairs of nodes
will increase restorative capacity, such additional
SUBSTITUTE SHEET (RULE 26)
CA 02276518 1999-06-30
WO 98/31159 PCT/US98/00058
- 21 -
links will undesirably increase underutilized link
capacity.
To partially offset underutilized link
capacity, spare links can also be used to reduce
traffic density. For example, if a particular working
link or group of working links becomes saturated, that
is, traffic is being transported at the maximum rate
and more traffic awaits to be transported, spare links,
if available, could be used to transport the additional
1o traffic. Such situations could occur in peak demand
situations, such as, for example, in a telephone
network on Mother's Day when there is typically a
significant increase in the number of calls. This
flexibility improves network performance and further
t5 reduces underutilized link capacity.
Although the embodiments described above
advantageously have decentralized control, the
reliability improvements provided by the allocation of
spare links or spare wavelengths are still applicable
2o even in those mesh networks with centralized control.
The restorative process will differ only in the time
needed to form alternative traffic paths. The process
of appropriate nodes activating appropriate switches to
connect spare links will be identical, thus giving the
25 centrally controlled network the same high restorative
capacity as decentralized networks.
Thus it is seen that a mesh network with
improved reliability is provided. One skilled in the
art will appreciate that the present invention can be
practiced by other than the described embodiments,
which are presented for purposes of illustration and
not of limitation, and the present invention is limited
only by the claims which follow.
SUBSTITUTE SHEET (RULE 26)
