Transcript Network and Service Discovery in Distributed Environments

Optical Networks,
Current and Future Technologies for WDM
By: Bogdan Ionescu
DSRG, University of Ottawa
July 5, 2000
1. Purpose
2. Introduction to WDM
3. Current Technology Standpoint
4. WDM Network Topologies and Architectures
5. Failure Resiliency
6. Switching and Routing
7. QoS in WDM
8. Access Networks
9. SONET and WDM
10. Control Plane for WDM Networks
11. Direction of WDM Networks
12. Conclusion
1. Purpose
•
•
•
•
Board study of WDM technology
Deployment
Work in progress
Establish future research areas in WDM
2. Introduction
•
Current optical signals
• have least signal loss around the region of 1300 or
1500 nm
• SONET use only one of these wavelengths at a time
•
WDM applies same principles as FDM to the optical domain
• many optical signals
• each with their own wavelength (carrier freq.)
• placed onto one fiber
2. Introduction
•
How WDM works?
• advances in optical components allow a window of
wavelengths around the 1300 or 1500nm thresholds
• with latest lab technologies
• window size of 100 wavelengths
• for OC-192 (9953.28 Mb/s) => 1 Tb/s
• this is DWDM
•
e.g. of such optical components [1]:
• Distributed Feedback Lasers (DFB)
• Erbium-doped Fiber Amplifiers (EDFA’s)
• Photo-detectors
• Multi-frequency Lasers
2. Introduction
•
Optical Components for WDM [1]
Component
Tuned Lasers
Multi-frequency Lasers
Tunable Filters
•
Purpose
Signal Generation
Signal Generation, Mux/Dmux, Routing
Circuit and Packet Switching, used in WR
Important Parameters:
• Tuning Time: time required for a laser/filter to focus
on a particular wavelength
• Tuning Range: spectral window, i.e. how many
wavelengths
2. Introduction
•
Properties of tunable lasers [1]
Source Category
Tuning Range
Tuning Time
Mechanically tuned lasers
10 – 20 nm
100 – 500 ms
Acousto-optically and electrooptically tuned lasers
10 – 20 nm
Tens of s
Good for packet switching
Injection current tuned lasers
4 nm
0.5 – 10 ns
e.g. limited tuning range
Switched
< 65 ps
Array sources (using AWG)
Typically 16
channels spaced by
200 GHz
100 – 200 ms
Locked  and temperature tuning of
whole comb by 275 GHz
Array sources (DFB)
Limited by number
of elements in array
1 – 10 ns
DFB lasers act independently, and can
drift to cause crosstalk
Switched Sources
Comments
Mechanically tuning of the facets FP
external cavity
Integration improves functionality and
speed
2. Introduction
•
Characteristics of tunable filters [1]
Filter Category
Tuning Range
Tuning Time
Comments
Fabry-Perot
500 nm
1 – 10 ms
Fiber implementation available
Acousto-optic
250 nm
10 s
Can be used as a router as well
AWG tunable filter
40 nm
10 ms
Thermo-optic tuning
Liquid crystal Fabry-Perot
30 nm
0.5 – 10 s
Electro-optic
16 nm
1 – 10 ns
Resolvable channels about 10
Fiber Bragg grating
10 nm
1 – 10 ms
Temperature or mechanical stretched
tuning
Cascaded Mach-Zehnder
interferometer filters
4 nm
50 ns
Mostly used for  conversion
Tunable filters based on
semiconductor or laser structure
5 nm
0.1 – 1 ns
Limited number of channels
Lower power consumption (< 1mW)
2. Introduction
(WDM Network Components)
•
•
Optical Add/Drop Modules (OADM)
• signal grooming and splitting
• usually made of Arrayed-Waveguide Gratings
(AWG) or Fiber Bragg Gratings (FBG)
 Crossconnect (WXC)
• key component in constructing networks
• composed of:
• Mux, Dmux, switches,  Crossconnect
2. Introduction
(WDM Network Components)
•
Reconfigurable  routing switches
a) employing  converters b) employing photonic switches
3. Current Technology Standpoint
•
Today WDM technology is mainly deployed for increasing
“pipe” capacity of backbone networks.
•
•
•
•
•
•
SONET interfaced with WDM Mux/Dmux
used for point-to-point connections
up to 100 times the wavelengths over same optical fiber
no need to lay new fiber for multiplied capacity on
backbone trunks
no known deployment of switching out routing WDM
(in development only)
WDM Mux/Dmux simple device requires minimal
control mechanisms for focusing lasers and filters.
4. WDM Network Topologies and Architectures
•
Broadcast and Select [1]
• sender relays information to all other nodes
• receiver node selects information of interest by use of
“enabling technology)
• e.g. a tunable filter to select the  carrying info.
• tunable transmitters (TT), tunable receivers (TR),
fixed transmitters (FT), fixed receivers (FR)
• Topologies
• star coupler, (N/2)log2N individual couplers
• bus, 2N couplers, high signal loss
• ring, high power dissipation better failureresilience
• MAC protocols needed
4. WDM Network Topologies and Architectures
•
Broadcast and Select Topologies [1]
4. WDM Network Topologies and Architectures
•
Broadcast and Select demonstrators: [1]
• LAMBDANET [16-20] late 80’s and early 90’s, one of
first, built by Bell Core
• 18  separated by 2nm & modulated @ 1.5 Gb/s
over 57 km
• RAINBOW I & II [1] in the early 90’s by IBM
• 32  separated by 1nm giving 300 Mb/s in I and
1Gb/s in II
• Other:
• STARNET I &II by Stanford University
• European Research and Development for
Advanced Communications in Europe (RACE)
4. WDM Network Topologies and Architectures
•
 Routing Networks [15,17]
• information is routed, switched and forwarded based
on 
• therefore obtain a physical and logical topology
•  reuse throughout network => increased capacity
• currently 4 to 32  networks exist and expected to
increase to 100 (DWDM) [15]
• virtual topology gives an optical layer to serve higher
layers e.g. offering VPN’s
• issues to consider:
• transparency, reliability,  reuse, virtual topology,
static/dynamic routing  assignment (RWA)
[15,17,18]
4. WDM Network Topologies and Architectures
•
 Routing Networks
4. WDM Network Topologies and Architectures
•
 Routing Networks demonstrators [1]
• Multiwavelength Optical Network (MONET) program
• developed by Bellcore funded by DARPA [18,19]
• uses 8  spaced by 200 GHz & modulated @ 2.5
Gb/s, in a ring topology
• latest demonstrators gave 10 Gb/s for 2000 km
• All-Optical Network (AON) between MIT and DCE
[20]
• uses 20  spaced by 50 GHz (0.4nm)
• Multiwavelength Transport Network by RACE [18]
• uses 4  spaced by 500 GHz & modulated @ 622
Mb/s or 2.5 Gb/s
• Other: WSAPNET, OPEN, MOSAIC, OPERA [18,16]
4. WDM Network Topologies and Architectures
•
WDM Ring [5]
• conceptually similar to a SONET ring
• main difference is in failure-resiliency
• Wavelength Cross Connect (WXC) interconnected in a
ring topology
• WXC use  switching and conversion
• allows for better service protection than SONET
•  (lightpath) protection in addition to path
and line protection.
• i.e. laser failure is fixed by converting , no
need for rerouting => faster recovery
• flexible and higher routing capacities
• [8] identifies eight types of WDM rings
• [25] studies several failure-resilient WDM rings
4. WDM Network Topologies and Architectures
•
WDM Mesh Networks [5]
• same optical components as for WDM rings i.e. WXC’s
• protocols are more complex
• failure-resiliency [9]
• routing &  assignment [10]
• likely to be used for backbone infrastructures
4. WDM Network Topologies and Architectures
•
Example of interconnected WDM network infrastructure;
Mesh, Ring, and Access ( Discussed Later )
5. Failure Resiliency
(Currently in SONET)
•
SONET/SDH survivable architectures [1]
• Automatic Protection Switch (APS) protocol
• protects against fiber cuts by automatically redirecting
traffic
• 1+1 protection:
• same signal is transmitted through two nonintersecting paths
• destination decides on which signal to choose
• 1:1 protection:
• two non intersecting paths; working and reserved
• reserved path is idle or used by low priority traffic
in “non-breakdown” times
5. Failure Resiliency
(Currently in WDM )
•
•
WDM networks will likely mirror SONET/SDH survivable
architectures.
Two type of classifications [11-14]
• Protection
• protection resource are reserved ahead of time
• protection resource may be used by low priority
traffic at “non-breakdown” times
• Restoration
• alternate routes of spare resources are discovered
at failure detection time
• if discovery of alternates fails, data is lost
• better resource utilization at “non-breakdown
times”
• slower than protection scheme (reconfig.. path)
5. Failure Resiliency
(Currently in WDM )
•
Two main schemes extensively implemented
•
First scheme: Fiber Protection Switching
• applied to point-to-point and ring architectures
• working fiber is backed up by another disjointed fiber
• entire fiber is switched
• granularity of one fiber
5. Failure Resiliency
•
(Currently in WDM )
Second scheme: At the Wavelength Level [6]
• applied to cross-connected mesh architectures
• slower but more bandwidth efficient
• Three variants:
• link
• fastest failure detection
• recovery is local to one point (i.e. @ link level)
• path
• reroutes between two end nodes
• full advantage of network spare capacity
• slower to detect and recover
• disjoint-link path
• at path setup time, an alternate path is
selected as well
• traffic restored immediately
5. Failure Resiliency
(IP over WDM)
•
For IP over WDM/DWDM
•
abolishes current architecture
• IP over ATM over SONET over WDM
• functional overlap
• slow to scale
• failure resiliency at different layer may
interfere with each other and provide network
instability
5. Failure Resiliency
(IP over WDM)
•
Detection and restoration at different layers
5. Failure Resiliency
(IP over WDM)
•
Proposed joint protection/restoration scheme coordinated at
both IP and WDM layers [6]
• needs traffic engineering for WDM networks
•
•
Optical control framework needed at optical layer
Key is combining [21]:
• recent advances in IP-MPLS-based control plane
constructs
• with OXC technology
• this is proposed by MPS [21]
• sharing of information between IP & WDM layers
• IP to directly access WDM channels
5. Failure Resiliency
(IP over WDM)
•
Proposed architecture for MPLS-based traffic engineering
[6]
• automatic topology discovery
• OXC to dynamically learn networks
• no manual intervention
• state information dissemination
• extension to link-state routing, e.g. OSPF
• link attributes added to state tables and
advertisement
• max. available link BW
• max. reservable link BW
• current BW reservation
• current BW usage
5. Failure Resiliency
(IP over WDM)
•
•
•
path selection
• RWA problem
• which LSP (Label Switched Path) to choose for
required  QoS values
• choose a centralized or distributed algorithm
path management
• establish, maintain, tear-down of LSP
• use of signaling protocols such as
• RSVP
• CR-LDP (Constraint-Based Routed Label
Distributed Protocol)
This allows for Joint Protection/Restoration at the IP/WDM
Layers
6. Switching and Routing
•
Have seen so far the physical and control mechanisms
required for switching / routing in WDM networks
•
Primary problem: Contention Resolution
•
Contention in WDM : when packets are switched and two
or more packets trying to leave the same port on the same 
•
Three types of contention resolution mechanisms [4]:
• optical buffering
• deflection routing
•  conversion
6. Switching and Routing
•
Optical Buffering (deflection in time) [4]
• delay lines
• fixed fiber lines each giving pre-calculated delays
• delays in term of packet durations: 1,2,3 ..., m
• packet sent directly to output or to required delay
line based on required delay
• recirculation fiber loops
• based on fiber loop delay line with multiple 
channels
• at contention, packet is converted to  available in
the loop
• kept circulating by corresponding passive fixed
filter
• converted to  available at output when
contention is resolved
• both suffer from increased signal to noise ratios
6. Switching and Routing
•
Deflection Routing (deflection in space) [4]
• one packet routed along desired “minimum distancerouting” link
• others are forwarded to greater than “minimum
distance-routing” links
• (optical buffer may still be used here)
• source-destination pair routes are no longer fixed, i.e.
different routes possible as in IP
• parameters that affect performance of deflection
routing
• network diameter (max. # of hops)
• deflection cost (increase in hops)
• problems similar to IP routing can arise esp. in
asynchronous transmission networks
• wondering packets, throughput collapse
6. Switching and Routing
•
 Conversion (deflection in  domain) [4]
•
•
•
convert additional packets to a different  following
the same next hop link as original packet
needs mechanism for choosing alternate  but on same
output fiber
best results were obtained if combined with buffering
• reduced probability of signal loss in buffering
6. Switching and Routing
•
General Comparison
• Buffering
• better network throughput at higher hardware
and control cost
• Deflection Routing
• easier to implement, cannot offer ideal network
performance
•  Conversion
• in-between with less signal loss
7. QoS in WDM
•
Differentiated Optical Services (DoS) [2]
• Provide Classes of Optical Services (DiffServ)based on:
• lightpath characteristics
• jitter
•  wander
• crosstalk
• amplified spontaneous emission
• available failure-resiliency characteristics
• protection
• restoration
• signal regeneration
• 3R, retiming and reshaping (signal
dependant), non-transparent but re-clocking
• 2R, regeneration, no retiming => jitter
• 1R no retiming and reshaping, simplest and
most transparent to optical format
7. QoS in WDM
•
DoS set of parameters characterize quality and impairments
of optical link
• quantitative: delay, jitter, BER, BW
• functional: monitoring, protection, security
•
Need for Classification and Mapping of Optical Service
• grouping OCh into classes reflecting previous QoS
parameters
• mapping aggregate DiffServ flows into corresponding
OCh grouping
• admission ctrl. and policing to ensure that ingress
DiffServ flows do not exhaust available optical
resources
• Architecture involves edge-devices performing the above
functions with a simple WDM core
• Need for monitoring & control plane to core devices
7. QoS in WDM
•
E.G. : Optical Service Classification
Criteria
Class 1
(1, 2)
Class 2
(3, 4)
Class 3
(5, 6)
BER
10-9
10-7
10-5
Survivability
90 %
70 %
20 %
Fine Grained
Coarse
Grained
N/A
Secure
Unsecured
Unsecured
3R
2R
1R
Monitoring
Security
Provisioning
7. QoS in WDM
•
General Mechanisms to enforce QoS:
• optical buffering according to priorities/classes
• deflection routing according to priorities/classes
•  conversion according to priorities/classes
• MPLS/OSPF/RSVP
•
Monitoring by way of:
• OADM’s (Optical Add /Drop Modules
• or opto-electric-opto conversion
8. Access Networks
•
Passive Optical Networks (PON) [5]
• placed at WDM Ring access level
• a bus or star topology
• medium access protocol (MAC) coordinates
transmission amongst users
• are less expensive than WDM rings due to lack of
active components
•
•
SONET rings
Room for QoS here as well:
• Dynamic Resource Allocation for Quality of Service on
a PON with Home networks [7]
8. Access Networks
•
Example of current Access Networks for WDM
9. SONET and WDM
•
•
•
•
10 years to finish migration to SONET [5]
Large investment made in SONET equipment
• North American Total as of 1998, $4.5 billion
For WDM to be accepted readily it has to integrate/be
backward compatible with SONET
Optical Transport Networks (OTN) [1]
• switched/routed WDM networks (OC-48 to OC-192)
• composed of WXC nodes
• control system for setup and teardown of lightpaths,
monitoring, and fault recovery
• needs definition of frame format (currently being
defined [22])
• OCh frame format to borrow SONET concepts
9. SONET and WDM
•
SONET Frame must be easily encapsulated into OCh frame
• e.g. OC-48 SONET frame cannot fit in an OCh
payload at OC-48 due to required OCh overhead
bytes.
• Some WDM equipment must have physical SONET
interfaces as specified in [23]
• SONET side need not be aware of WDM
• WDM equipment might have to do  conversion at
SONET interface
• APS of WDM must not interfere with that of SONET
• must react faster than SONET (50 ms)
• recovery at lower level (WDM) 1st
• Recall: WDM restoration is faster but detection is
slower
9. SONET and WDM
•
E.G. on WDM ring a WXC can directly drop and add into
the optical medium the  used in the SONET ring
•
Organizations dealing with interoperability issues
• Optical Internetworking Forum (www.oiforum.com)
• ITU-T Study Group 15 (www.itu.int/ITUT/com15/index.html)
• SONET Interoperability Forum
(www.atis.org/atis/sif/sifhom.htm)
10. Control Plane for WDM Switches/Routers
•
WDM OTN networks need to support
• APS
• QoS
• Monitoring
• Security
• Interoperability (SONET, PON, ....)
•
Each device (edge or core) needs to support some or all
points above
• need to define frame format
• signaling protocols (setup and teardown of  paths)
• MIB’s
• management protocols
10. Control Plane for WDM Switches/Routers
•
•
•
•
•
A uniform control plane that is distributed, supporting
heterogeneous environments is needed
Recall: SONET suffers immensely from lack of a
standardized management platform
Ongoing effort to integrate, control, and manage existing
networks in a vendor independent way [24]
CORBA’s Xo/JIDM specification is a foundation [5]
• joint effort by major Telco. Industries
• integrates management and control architectures for
• ATM, SONET, WDM, IP, and others
This allows WDM networks to fully support OAM&P
• i.e. operations, administration, maintenance, and
provisioning
11. Direction of WDM Networks
•
•
•
•
More than just the current point-to-point “fat-pipe”
capacity
IP over WDM/DWM for OTN support
• OSPF/MPLS/RSVP
• increase in: control, monitoring, & management
Support for QoS reaching into core
Advances in tuned lasers, filters,  converters
• increase in tuning speed and range
• increase in  window along base ’s (1300,1500 nm)
• allow WDM to become circuit/packet switched
networks offering value-added features opening up
diverse IP and WDM services to the paying public
12. Conclusion
•
•
•
•
Currently WDM applied in increasing point-to-point
capacity
Increasing towards providing WDM OTN with value added
features
WDM / SONET Integration
Many critical components missing:
• IP over WDM protocol integration
• Control Plane
• Traffic monitoring and measurement
• Traffic engineering
• QoS developed leveraging above protocols
• WDM access networks supporting QoS
13. References
[1] J. Elmirghani and H. Mouftah, “Technologies and Architectures for Scalable Dynamic Dense WDM Networks,”
IEEE Commun. Mag., vol. 38, no. 2, February 2000, pp. 58-66.
[2] N. Golmie, T. Ndousse and D. Su, “A Differentiated Optical Services Model for WDM Networks,” IEEE
Commun. Mag., vol. 38, no. 2, February 2000, pp. 68-73.
[3] I. Van de Voorde, and C. Martin, “The SuperPON Demonstrator: An Exploration of Possible Evolution Paths for
Optical Networks,” IEEE Commun. Mag., vol. 38, no. 2, February 2000, pp. 74-82.
[4] S. Yao and B. Mukherjee, “Advances in Photonic Packet Switching: An Overview,” IEEE Commun. Mag., vol.
38, no. 2, February 2000, pp. 84-94.
[5] D. Cavendish, “Evolution of Optical Transport Technologies: From SONET/SDH to WDM” IEEE Commun.
Mag., vol. 38, no. 6, June 2000, pp. 164-172.
[6] Y. Ye, S. Dixit and M. Ali, “On Joint Protection/Restoration in IP-Centric DWDM-Based Optical Networks”
IEEE Commun. Mag., vol. 38, no. 6, June 2000, pp. 174-183.
[7] J. Jang and E. Park, “On Joint Protection/Restoration in IP-Centric DWDM-Based Optical Networks” IEEE
Commun. Mag., vol. 38, no. 6, June 2000, pp. 174-183.
[8] S. Johansson et al., "A Cost-Effective Approach to Introduce an Optical WDM Network in the Metropolitan
Environment," IEEEJSAC, vol. 16, no. 7, Sept. 1998, pp. 1109-22.
[9] S. Ramamurthy and B. Mukherjee, "Survivable WDM Mesh Networks, Part I - Protection," Proc. INFOCOM'99,
Vol. 2, 1999, pp. 744-51.
[10] A. Mokhtar and M. Aziziglu, "Adaptive Wavelength Routing in All-Optical Networks," IEEE/ACM Trans. Net.,
vol. 6, no. 2,Apr. 1998, pp. 197-206.
13. References
[11] P. Demeesteretal., "Resilience in Multilayer Networks," IEEE Commun. Mag., Aug. 1999, vol. 37, no. 8, pp.
70-76.
[12] S. Shew, "Fast Restoration of MPLS Label Switched Paths," Internet draft, work in progress, Oct. 1999.
[13] S. Makam et al., "Protection/Restoration of MPLS Networks," Internet draft, work in progress, June 1999.
[14] M. Medard, S. Finn, and R. Barry, "WDM Loop-back Recovery in Mesh Network," Proc. IEEE INFOCOM '99,
vol. 2, pp. 752-59.
[15] D. J. Blumenthal et al., "Special Issue on Photonic Packet Switching Technologies, Techniques and Systems,"
1EEE/OSA J. Lightwave Tech., vol. 16, no. 12, Dec. 1998.
[16] H. T. Mouftah and J. M. H. Elmirghani, “Photonic Switching Technology-Systems and Networks”, IEEE Press,
July 1998.
[17] R. L. Cruz et al. (Eds.), "Special Issue on Optical Networks," IEEE JSAC, vol. 14, no. 5, June 1996.
[18] M. Fujiwara et al. (Eds.), "Special issue on Multiwavelength Optical Technology and Networks," IEEEIOSA J.
Lightwave Tech., vol. 14, no. 6, June 1996.
[19] M. Wang, Ed., Feature Topic on Multiwavelength Fiber Optic Communication, IEEE Commun. Mag., vol. 36,
no. 12, Dec. 1998.
[20] D. A. Smith et al., "Evolution of the Acousto-Optic Wavelength Routing Switch," 1EEE1OSA J. Lightwave
Tech., vol. 14, no. 6, June 1996, pp. 1005-19.
13. References
[21] D. Awduche et al., "Multiprotocol Lambda Switching," Internet draft, work in progress, Nov. 1999.
[22] ITU-T Rec. G.709, "Network Node Interface for the Optical Transport Networks," work in progress, SG 15.
[23] ITU-T Rec. G.957, "Optical Interfaces for Equipments and Systems Relating to the Synchronous Digital
Hierarchy," July 1995.
[24] R. Pease, "Companies Demonstrate Multivendor, Multitechnology Network Management," Lightwave, Nov.
1999.
[25] J. Manchester, P. Bonenfant, and C. Newton, "The Evolution of Transport Network Survivability," IEEE
Commun. Mag., vol. 37, no. 8, Aug. 1999, pp. 44-57.