1

Short repetition of two important facts (1)

11 11

11 1111 1144 44

11

22

33

44

11

22

33

44

Tidligere utbygging

RegeneratorTerminalFiber

Før: 1 kanal pr fiber

Optiskforsterker

MultiplekserDemultiplekser

2,5 Gb/s =30000

Opptil20 000 000

WDM: 4-128 kanaler

pr fiberNåværende utbygging

Wavelength Division Multiplexing(WDM), mangedobler kapasitet i fiber

Electronic/electrooptical

Now Optical amplifier

WDM: 4-128 channelspr fiber

1 channel pr fiber

Up to

Earlier

2

Optical switches• Circuit switching

– Switches signals between fibers and/or wavelengths.

• Wavelength conversion– To avoid collision on wavelengths (in same fiber)

Optisk krysskopler

BølgelengdeKonverter

I1

I2

I3

I4

U1

U2

U3

U4Wavelengthconverter

Optical crossconnect

3

Repetition Outline:

Optical componentsTransmission aspectsOptical Transport Network –OTN- G. 709Optical protection switchingOPS/OBS

4

Optical fibre, characteristic

• Large bandwidth (theoretical 50 THZ)• Low attenuation (0,2 dB/km at 1550nm).• Physical size beneficial, light and thin, simplifies

installation • Splicing and mounting connectors more complex• Immune to electromagnetic interference• Environmentally friendly material (sand!).

5

Propagation through fibre

• Lightpulses are reflected into the core when hitting the cladding => approximately zero loss

Andreas Kimsås, Optiske Nett

6

Coupling light into the fibre

• Single modus– Coupling into the tiny 10 micrometer core is demanding

– Lining up the light-source is a significant part of the production cost

• Multimode– Larger core diameter simplifies coupling

7

Modulation

• OOK modulation (on-off-keying) – NRZ (No Return Zero) most often used– RZ (Return Zero), some use– Phase modulation currently not popular– More advanced modulation formats being launched for 40 Gb/s pr.

Channel systems.

• Directly modulated laser– For medium bitrates Gigabit

• External modulation, e.g. Employing external modulator: MZ interferometer

– For high bitrates 10 Gb/s and beyond

8

What is a long distance?

• 100 m?– LAN

• 10 Km?– Access network

• 1000 Km?– Transport network

9

Long distance optical system

• Attenuation must be compensated– Regeneration

– Attenuation

• Dispersion must be compensated– Dispersion compensation employing fibre

– Electronic compensation

10

Regeneration• 1R regeneration = Amplification (Reamplification)

– Amplifies the signal without conversion to electrical – Typically transparent for signal (shape, format and modulation)– Usually an optical amplifier

• 2R Reamplification & Reshaping:– Reshapes the flanks of the pulse as well as the floor and roof of the pulse, removes noise. – Usually electronic– Optical solutions still subject to research

• 3R Reamplification & Reshaping & Retiming:– Synchronisation to original bit-timing. (regeneration of clock)– Usually involves electro-optic conversion – Optical techniques in the research lab.

11

Erbium Doped Fiber Amplifier (EDFA)• Widely deployed in optical networks

12

Available wavelength range depends on amplifier technology

0

0,1

0,2

0,3

0,4

0,5

1200 1300 1400 1500 1600

Wavelength (nm)

Lo

ss (

dB

/km

)

EDFAEDFAC - bandC - band

1530-15621530-1562

EDFAEDFAL - bandL - band

1570-16001570-1600

ALTERNATIVE AMPLIFIER TECHNLOGIES: RAMAN AND SOA

PDFAPDFA1300 nm1300 nm

Commercially available Still subject to research

13

Dispersion• Pulse spreading when propagating through the fibre.• To much spreading results in intersymbol- interference• Limits the maximum transmissionrate through the fibre. • Three types of dispersion:

– Modi-dispersion: Light traveling in different modi undergoes different delays through the fibre. Not present in SM!

– Material-dispersion (chromatic): Refractive index is function of wavelength– Waveguide-dispersion: Propagation of different wavelengths depends on

the characteristic of the waveguide, e.g. Index, geometry of core and cladding.

14

Zero dispersion

• At 1300 nm in standard fibre– Material (chromatic) dispersion is close to zero at 1300 nm

– Not minimum loss

• ~ 1500 nm in dispersion shifted fibre– Manufactured for zero dispersion in 1500 nm region

– Design core and cladding to give negative waveguide dispersion

– At a specific wavelength, material and waveguide dispersion will result in zero total dispersion.

15

Chromatic dispersion

Figure: S. Bigo, Alcatel: Talk at Norwegian electro-optics meeting2004

16

Chromatic Dispersion in transmission fibre – key figures• Dispersion depends on fibretype • G652, “Standard fibre” -17 Ps/nm*km @ 1550 nm• Dispersion shifted fibre: 0 dispersion @ 1550 nm• Non – Zero (NZ) dispersion shifted fibre: -3 to -6

Ps/nm*km

17

Dispersion Compensating Fibre (DCF)• Negative dispersion compared to transmission fibre• Much higher dispersion/km => Shorter fibre than

transmission fibre required for achieving zero dispersion

18

Long distance fibre-optical transmission Basic optical components(For WDM systems: Additional mux/dmux required

Transmitter(Laser+

modulator)

Receiver(fotodiode +

amplifier)Long Fibre

Compensation of amplitude and dispersion

EDFADCF

19

Optical couplers: splitter & combiner• One or more fibers in, several fibres out

– Divides the optical signal on several fibres.

• Signal power is divided on the output-fibres• Splitting ratio is varying

– 50/50, 50 % on each of two fibres– 10/90, 10 % in one, 90 % in a second.

• Attenuation from input to output depends on splitting ratio– 50/50 splitter results in 3 dB attenuation (halving the power)

Combiner

Splitter

20

Arrayed waveguide Grating• 1 X N or N X N coupler divides the light on N

waveguides of different length• Waveguides is then coupled together, resulting in

interference • On each of the N outputs, constructive interference is

achieved for a specific wavelength and destructive interference for the other wavelengths

21

Multiplexing/Demultiplexing

• Optical multiplexing: Couple several waveguides together into a fibre.

• Optical demultiplexing: Separate wavelengths from an input fibre into several output fibres with a single wavelength in each.

• Employed for mux/demux in optical network nodes

22

Optical add/drop• Filter out a wavelength or a set of wavelengths. • Does not employ non-linear effects• Add a wavelength or set of wavelengths. • May be reconfigurable (ROADM)

– Select which wavelength to drop and add on a fibre

– Not as configurable as an optical cross connect (full cross-connection between several fibres)

1,2 1,2

11

Drop Add

23

Most important components• Optical fibre

– Principle– Key parameters: Chromatic Dispersion, attenuation

• Optical coupler– Principle– Application

• EDFA Optical Amplifier– Principle– Application

• AWG– Principle– Applications

24

Coarse WDM• Cheaper technology with less scalability than DWDM• Typically maximum 16 channels

0

0,1

0,2

0,3

0,4

0,5

1200 1300 1400 1500 1600

Wavelength (nm)

Lo

ss (

dB

/km

)

2 dB/km

G.652G.652C

1271-1451 nm 1471-1611 nm

25

16 channel CWDM using two multiplexers for two different bands

EXT

EXT

C1 – (8+1) C1– (8+1)

C1 – 8L C1 – 8L

14711491151115311551157115911611

14711491151115311551157115911611

12711291131113311351137114311451

12711291131113311351137114311451

26

CWDM and DWDM hybridC1-8

D1-52 D1-52

14711491151115311551157115911611

C1–8

1535.82 1535.04 1534.25 1533.47 1532.68 1531.90 1531.12

1530.33

14711491151115311551157115911611

1535.82 1535.04 1534.25 1533.47 1532.68 1531.90 1531.12

1530.33

27

Optical networks (Zouganeli)• Increased traffic demands (e.g. from broadband home

users/businesses and new services) => Fat pipes needed.• ”IP everywhere” and development in optical technology => Fokus on

simplifications:

28

Reconfigurable (R-)OADM• Still use cross connect for some wavelength/wavebands, but

introduce more flexible add-drop function:

Not singlewavelength!

29

Network element functionality: Bypass

• Traffic bypassing intermediate IP routers => Less load on routers (can be smaller and cheaper)

• In meshed networks:Used to directly connect node pairs with high traffic between them.

• (UNINETT is in the process of doing this now).

30

Transparent (all-optical) switches (1)

• Micro-electro-machining systems (MEMS)

• Complicated, but has received a lot of attention.

31

Transparent (all-optical) switches (2)

• Probably most promising alternative currently.

• However: Tunable Wavelength Converters (TWCs) are very expensive.

32

Switching architectures with wavelength conversion (Borella)

• Dedicated converters for each output – Many converters

– Flexible, no blocking

– Wavelength specific multiplexers minimizes attenuation.

33

Switches with shared wavelength conversion (Borella)

• Shared between all input lines– Access from any input wavelength

– Optimal wavelength converter resource utilization

– WC may not be available if too few

– Extra switch between WC and output MUX required.

34

Needed functionality for optical OXC based networks (1)• Opto-electronic or all-optical.• Scalability and flexibility

– Handles much higher number of line ports and directions than R-OADM– Higher flexibility than R-OADM

• Service provisioning: End-to-end lightpaths should be provisioned in an automated fashion (not necessarily all-optical or same wavelength end-to-end).

• Protection and restoration: Must have mechanisms to protect against fiber cuts or equipment failure at nodes. I.e. redirect traffic from failed to backup paths.

• Wavelength conversion: Lightpaths can change wavelength to increase flexibility in allocating network resources. Much easier to implement in opto-electronic OXC than in all-optical OXC;3R versus 2R (Mach-Zhender interferometer).

35

• Multiplexing and grooming: Normally done in the opto-electronical add-drop part.

• Today mainly opto-electronic solutions.• Many candidate all-optical solutions:

- Generic switch architectures (Clos, Shuffle,..) where elements are simple optical switch elements, connected with fibers. - ”Broadcast and select” switching matrixes realized with splitters and Semiconductor Optical Amplifiers (SOAs) (0 – 1 : block or let-through light).- Two- or three dimensional array of micro mirrors (MEMS)- Tunable wavelength converters and Array Waveguide Gratings (AWG)

Needed functionality for optical OXC based networks (2)

36

4 different architectures1a) Fixed patch panel between WDM systems with transponders.

1b) Electrical switch fabric between WDM systems with transponders.

1c) Transparent switch between WDM systems with transponders, complemented by a OEO switch for drop traffic.

1d) Transparent switch on a transparent network. The signal stays optical until it exits the network.

37

1c) Transparent switch between WDM systems with transponders, complemented by a OEO switch for control and management functions

The optical switch fabric is bit-rate independent and accommodates any data rates available. Most lightpaths will bypass the OEO switch.

The drop side ports are connected to an OEO switch that provides SONET/SDH line termination through its opaque ports.

38

Optical Transport Network (OTN)

ITU-T standard G.709Paper: Andreas Schubert: ”G.709 – The Optical

Transport Network (OTN)”

http://www.jdsu.com/product-literature/g709otn_wp_opt_tm_ae.pdf

39

Why OTN?

• Standard for optical networks required– Optical interconnection between equipment from different vendors– Optical interconnection between different operators

• Once called “digital wrapper”• Framing of different protocols for transport over the physical

optical layer– E.g. IP/Ethernet or IP/ATM or SDH

• Takes SDH/SONET further, enabling optical functionality – Six level Tandem connection monitoring (TCM)– From a single to multiple wavelengths– Forward Error Correction (FEC)– What is FEC?

40

OTN hierarchy Client

OH Client

OH OPUk

OH ODUk FEC

OCh payload

Client

OPU

ODU

OTU

OChannel

OMS payload

OTS payload

OCCp OCCp OCCp OCCp OCCp OCC

Non

A

ssoc

iate

d ov

erhe

ad

41

ODU OH

• PM - Path Monitoring, contains three sub-fields• TCM1-TCM6

– OH for six independent TCM’s– Contains similar sub-fields as PM

• TCM/ACT Activation/deactivation of TCM• GCC

– Communication between network elements (management), two channels

• APS/PCC– Automatic Protection Switching– Protection Communication Channel

• RES Reserved for future use• EXP Experimental use• FTFL - Fault Type and fault location Channel

– Fault status, type and location– Related to TCM span

RES TCM/ACT FTFLTCM6 TCM5 TCM4

TCM3 TCM2 TCM1 PM EXP

RESAPS/PCCGCC2GCC1

42

FEC algorithms• Performance increase depends on algorithm and

amount of overhead (redundancy information)• Standardized algorithms, G.709. Reed-Solomon

based

43

Summary OTN

• Management to the high bandwidth WDM network– SDH/SONET single wavelength, OTN – multiple wavelengths

– Builds on management functionality from SDH/SONET

– Monitoring functionality

– GCC channels for management communication

• Transparency to other protocols, e.g. IP– Wrap whatever you like

• FEC compensates physical impairments, increases cost-efficiency

44

Summary OTN

• Management to the high bandwidth WDM network– SDH/SONET single wavelength, OTN – multiple wavelengths– Builds on management functionality from SDH/SONET– Monitoring functionality– GCC channels for management communication

• Transparency to other protocols, e.g. IP– Wrap whatever you like

• FEC compensates physical impairments, increases cost-efficiency

• OTN switching is being deployed• OTN transmission and switching market is increasing rapidly

45

Protection in mesh network (Also used in other networks)

• 1+1 protection – Continuous signal on two alternative paths, choose the best.

– Hitless protection switching possible (switching without loss)

• 1:N protection– Several parties share a single common protection path.

– Enables the path to be employed by low priority traffic when not in use.

– Implies information loss because of switching. Data in the fibre is lost. Not hitless.

46

Protection schemes

• Dedicating resources– Dedicated protection (e.g. separate dedicated wavelength)– 1+1 duplicating data, or 1:1, pre-empting low-pri. data– In context of rings these are called DPRings – Optical unidirectional path switched rings (OUPSRs), path– Optical unidirectional line-switched rings (OULSRs), line

• Shared protection– Protection resources shared between several lightpaths– 1:N, requires signalling– In context of rings these are called SPRrings– Optical bidirectional path switched rings (OBPSRs), path– Optical bidirectional line-switched rings (OBLSRs), line

47

Classification of resilience schemes WDM

• Restoration– Dynamic lookup for backup paths, spare capacity in the network.

Typical IP-layer.

• Protection– Reserving dedicated backup paths in advance. Typical WDM layer.

48

Protection times

• SONET/SDH– 60/50 milliseconds for establishing a connection.

– May avoid interruptions in a phone call.

• Optical layer– 2 micro – 60 milli seconds.

– SONET detects errors within 2.3 – 100 micro seconds. Protection at higher layers may be initiated.

• IP-layer– Slow detection, calculation and signalling, typically seconds.

• MPLS– Relatively fast detection with HELLO messages, the higher

frequency of the messages, the higher the overhead.

– Fast switching if pre-planned path, LSP.

49

Carrier-Grade Ethernet Technology

Based on the article:“Ethernet as a Carrier Grade Technology: Developments

and Innovations” by R. Sanchez, L. Raptis, K. Vaxenavakis

50

Native Ethernet characteristicsEthernet Frame: 1) 7octets Preamble for synchronization 2) SFD (10101011) start of

MAC frames

3) 48 bit DestinationAdress,

48 bit SourceAddress

Simplicity (plug n’play) and cost effective

The switching logic (self-configuration)

Listening, Learning and Forwarding

Redundancy through xSTP

VLAN known as a broadcast domain

Connection-less (single hop)

CSMA/CD (do we still need it in switched Ethernet?)

51

Why Carrier Ethernet ?• Ethernet is the technology of choice in the customer domain (85% of all

networks and 95% of all LANs)

• Internet is packet-switched, suitable to be transported over Ethernet

• Eliminate potential internetworking problems between the core (carrier) network and Ethernet acess networks.

• High bandwidth with simplicity and low cost?

The MEF1) has defined Carrier Ethernet as “an ubiquitous, standardized, carrier-class Service and Network defined by five attributes that distinguish Carrier Ethernet from familiar LAN based Ethernet”

Standardized services Scalability Reliability QoS Service Management

52

Carrier Ethernet Challenges

Moving Ethernet from the LAN to the carrier network brings out requirements/challenges:

1. Scalability

– Support for 10exp6 customers of an SP

– Evolving the VLAN-tagging standards

2. Protection (Reliability and Resiliency)

– Achieve the required 50ms recovery time

– Problems with xSTP recovery time, other protocols required

3. Hard QoS comparable with the guaranteed service from existing leased lines

4. Service Management

– Service provisioning based to SLAs

– Service Monitoring and Troubleshooting

5. TDM support (Inter-working with existing technologies)

53

Scalability- MAC-in-MAC

header encapsulation

-The MAC header is

added at the edge of the

SP

- 24 bit I-SID

~16 service instances

- Dedicated set of

MAC addresses

-Total separation of the

customer and SP networks

54

PBB-Traffic EngineeringPBB-TE 802.1Qay introduce connection-oriented forwarding mode and Ethernet tunnels:

Deterministic service delivery, QoS Resiliency OAM requirements Turning off xSTP Forwarding is not based on the MAC learning mechanism but provided by

the OAM plane

55

Operation, Administration and Maintenance (OAM)

Important building block toward carrier services Ethernet, multiple working/standardization bodies.

IEEE 802.1ag and ITU-T Y.1731: Fault detection through Continuity Check Messages Fault verification through Loopback and reply messages Fault Isolation through Linktrace and reply messages

ITU-T Y.1731 Fault notification through Alarm Indication Signal Performance monitoring

Frame Loss Ratio Frame Delay Frame Delay Variation

56

Conclusions

• Its simplicity and cost-effectiveness makes Ethernet a desirable technology

for the NGN carrier networks

– Can Ethernet still be considered ”simple” after the discussed

changes???

• Native Ethernet is lacking capabilities for the MAN and WAN environment.

• PBB, PBB-TE and OAM aim to enhance Ethernet and provide the required

carrier-grade services as from SONET/SDH, ATM and MPLS.

• Resiliency?

• Work in progress!

57

Access technologies properties: xDSL

• Typically asymmetric, downlink 1/4-1/8 of uplink• Twisted pair copper cable, fundamental physical limit

is close, Shannon theorem• Bandwidth/distance tradeoff

VDSL required for high capacity triple play

ADSL/RealADSL2

CapacityMbit/s

Distance (Km)1

6

25

6

31.5

15

ADSL

VDSL Shannon

52

58

•Consentrator => less fibers, needs power

•Many Fibers => no external power is needed

•Passive =>Higher power loss Do not need power

Fiber to the Home (FttH) variants

59

PON: SCMA, TDMA, WDMA

• Sub Carrier Multiple Access (SCMA)– Unique RF frekquency to each subscriber. Share wavelengths

• Time Division Multiple Access (TDMA)– Collision avoidance with access protocols

– ATM-PON (B-PON), Gigabit PON (G-PON), Ethernet-PON (E-PON), Gigabit Ethernet PON (GE-PON)

• Wavelength Division Multiple Access (WDMA)– no collisions

– higher capacity

– more expensive

60 Passive Optical Network (TDMA)

OLT

ONU

up to 20km

OLT: Optical Line Terminal ONU: Optical Network Unit

downstream

passive splitter

Limitation on power budget

Time-sharing offiber resources

Burst mode transmissionDifferent power from each subscriberMakes capacity upgrades difficult

61

passive splitter

upstream

Passive Optical Network (TDMA)

OLT

ONU

up to 20km

OLT: Optical Line Terminal ONU: Optical Network Unit

FttH architecture comparison

pros:

passive fibre plant

low OpEx

one connection at OLT

cons:

broadcast centric

less scalable

less upgradeable

complex customer differentiation

62

Downstream Ethernet-PON • ATM is expensive, Ethernet sells in high volume and

is therefore cheap– QoS og VLAN

• Fiber resources in E-PON is shared and Point-to-Point

• Ethernet broadcast downstream (as in CSMA/CD)– All frames are received by all subcribers

– Upstream the ONUs must share capacity and resources

63 Upstream and multiple access• Collisions must be avoided

– Too long distances implies a too long collision domain

• Time-sharing is therefore preferred, timeslots to each ONU

• All ONUs are synchronized to a common time-reference– Buffer in ONU assembles packets and sends in time-slot

– Allocation of resources is an issue

64

WDM PON for the future

• GPON/EPON may not handle future requirements on bitrate

• 10GPON – 10 Gb/s– Power budget imposes severe limitations on distances and splitting

ratio

• WDM-PONs solves the limitations of TDMA-PON– Dedicated wavelength to each subscriber– May be combined with TDMA-PON in a hybrid, allowing 1:1000

splitting ratio.– Many variants of WDM-PON

65

WDM, One wavelength to each subscriber

OLT

ONT

WDM-PON (WDMA)

66

Basic WDM-PON architectures• B&S architecture

– Passive splitter– Unique filter in ONU– Individual wavelength

upstream– Broadcast security issues

• AWG based– Low insertion loss, 5 dB– Universal Rx– Wavelength specific Tx– Periodic routing behavior

• AWG + Identical ONU’s– Single shared wavelength

upstream (TDMA)– Broadband LEDs and

spectral slicing give poor power budget

– Bidirectional OLT using a circulator

67

Most Cost effective: CWDM-PON • 16 CWDM wavelengths on SFW supports 8 ONU’s

– 1270 nm to 1610, ITU-T standard

• High power budget but potential problems with old fibers (OH peak)

• Employs standard low-cost pluggable SFP modules– Capex is low, Opex moderate (higher than colourless)

• DWDM much more expensive than CWDM, why?

68

Power budget CWDM• What is a power budget?• What is it useful for?• What causes the greatest loss?• Why is the power budget higher for DWDM compared to CWDM

69

CAPEX Cost on different PON-solutions• CWDM most cost-effective, but lowest splitting ratio• Amplified TDMA highest splitting ratio

70

Unified infrastructure: core to access• PON not only to

residentials• Mobile back-haul• ADSL back-haul• Enterprise networks• Combine with WDM

Metro rings• Combine with ROADM

nodes• Cost optimization

– Common management and control plane required

– Common protocols required (Not SDH and Ethernet and…)

71

Layered Network Management

Core Network

Metro SubNetwork 1Metro SubNetwork 2

EMS 2

Core EMS

NMS

O1

O2

O3

O4

O6O7

O8

O9

O10

M1M2

M3

M4

M5

M6

EMS 1

72

GMPLS Introduction• IP -> MPLS

=> Datagram to Virtual Connection (VC) (point-to-point)

• Explicitly routed label switched paths (LSPs) established before information transport – independent of actual routing paradigm

• Label swapping used as forwarding paradigm• Forwarding equivalence classes (FECs)• Label hierarchy / Label stacking

73

MPLS Label Forwarding Example

LABEL SWITCHINGIP Forwarding IP Forwarding

IPPacket

Label 1

IPPacket

IPPacket

Label 2

IPPacket

Label 3

IPPacket

Label-Switched Path (LSP)

LERLERLSRLSRLSRLSR

LERLER

74

75

Introduction (2)• Constraint based routing

- traffic engineering (QoS differentiation)- fast reroute (after failure)- diversity routing (disjoint alternative paths for protection)

• Routing protocols (e.g. OSPF) must exchange sufficient information for ”constraint”

• Resource reservation protocol with traffic engineering (RSVP-TE) is used to establish LSP/label forwarding states along path.(The alternative CR-LDP is not used any more)

76

Introduction (3)Generalized MPLS:• Extensions to handle e.g. optical network resources (OXC’s) (e.g.

extensions of OSPF, RSVP-TE).• Common control plane for packet and optical network• New Link Management Protocol (LMP) for optical links.• Support for (label) switching in time, wavelength and space domains

– and a label hierarchy. • Additional functionality to handle bidirectional links and

protection/restoration.

77

ER-LSP setup example using RSVP-TE

RSVP Path message carried EExplicit RRoute OObject (ERO)

RSVP Resv message carries Label information (L)

LSR8

LSR2

LSR6

LSR3

LSR4

LSR7LSR1

LSR5

LSR9

ERO=(2, 6, 7, 4, 5)

ERO=(6, 7, 4, 5)

ERO=(7, 4, 5)

ERO=(4, 5)

ERO=( 5)

L=21

L=10

L=21

L=14L=5

78

Enhancements to signalingBidirectional LSP setup (New in GMPLS):• Bidirectional optical LSPs (lightpaths) are important for network operators

– Fate sharing

– Protection and restoration

– Same QoS in both directions, same resource demands

Problems with two independent LSPs in MPLS:• Additional delay in set-up (problem in protection)• Race conditions for scarce resources => lower probability of success for both

directions simultaneously• Twice the control overhead

In GMPLS: Single set of Path/Request and Resv/Mapping messages used to establish LSPs in both directions at once.

79

Enhancements to signalingNotify messages:• Added to RSVP-TE for GMPLS • Provides a mechanism for informing nonadjacent nodes

of LSP-related failures.– Inform nodes responsible for restoring connection

– Avoid processing in intermediate nodes

• Speed up – Failure detection and reaction

– Re-establishment of normal operation

80

GMPLS Restoration

• When fault is handled after a failure has occurred – Dynamic resource allocation

• Usually at least one order of magnitude higher delay than protection

• Different levels of ”preparedness” – Pre-calculated routes or not;

– Some resources reserved or not

81

GMPLS Protection and Restoration (3)

Protection mechanisms:• 1+1 protection: simultaneous transmission of data on two different paths.

• M:N protection: M pre-allocated back-up paths shared by N connections. (1:N is most usual; 1:1 also relevant).

• Span protection – between adjacent nodes (NB! Avoid ”fate sharing”):

82

GMPLS Protection and Restoration (4)• 1+1 Path protection (disjoint paths):

• For M:N Path protection: back-up paths may be used for lower priority traffic in normal operation – preemption (Supported by GMPLS)

83

GMPLS Protection and Restoration (5)• Restoration mechanisms:

• Alternative paths may be computed beforehand, but resources are seldom allocated before they are needed.

84

TTM1: ”Burst, packet and hybrid switching in the optical core network”Steinar Bjørnstad et al.

85

IntroductionWanted:

High capacity optical layer network with following requirements:

• Support high utilization of resources

• Support fine granularity

• Support quality needed for strict real-time services

• Support variable length packets

86

Some more OBS ”reservation” schemes

Reserve a Fixed Duration(RFD)

- reservation knows burst start and end time

One-way burst reservation schemes(No transfer acknowledgement before sending data)

Just EnoughTime (JET)

In-Band Termination(IBT)

- burst contains burst"terminator"

Tell-And-Go(TAG)

- separate burst tear-down

Just-In-Time(JIT)

Horizon

Reserve a Limited Duration(RLD)

- reservation contains burstend time

LAUC LAUC-VFJumpStart

87

Packet/burst handling schemes

88

OpMiGua node design (Hybrid switch)

Buffer can be electronic (RAM) or based on FDLs (or some future invention?)

89

TTM1: OPS and OBS

“The Application of optical packet switching in Future Communication Networks”Mike J. O’Mahony et al.

90

Optical Packet Switching (OPS)• Bring packet switching into the optical domain• Switch packets at the optical layer• Fast optical switching matrixes required

– Nanosecond switching time

• Pure OPS still fare away– Packet header recognition, demonstrated for short headers

– Header generation, demonstrated hardware encoded headers

– Packet control (setup of switching matrix), far away …

– Transparent payload switching: Feasible

• Focus here is optical payload switching, but electronic header processing.

91 Adding OPS to an OCS/OXC network• A network that supports both SDH circuits and packet transport

– Some wavelengths are reserved for SDH connections only– Other wavelengths are used for packet transport

• Aggregation of packets in edge nodes (electronic RAM available for contention resolution in edges).

• Packet switch in core added (if not all packets aggregated must be sent to same destination …).

=> OCS and OPS shares same infrastructure, but is in reality two different logical networks.

If large networkmore core OPSSwitches addedfor better utilizationof capacity.

92 Aggregation of packets in edge nodes• Topological and logical interface between service

and transport layers.• Fast switching and packet traffic aggregation at

edge of network; dynamic and fast wavelength allocation needed.

• Only some wavelength channels allocated to packet traffic.

• OPS and IP domains can have integrated control plane.

• OPS needs additional information about the OTN (topology, configuration etc).

• Wavelength channels used for packets may terminate in other edge nodes or in a core node. In latter case a new wavelength channel is used to next (edge or core) node.

93

Optical memory• Large buffers are difficult to realize using FDLs only.• Suggested to use combination of electronic (RAM) and FDLs.• Electronic (RAM) to handle long delays.• Optical (FDL) to handle short delays.• Simulation: Probability that a randomly chosen byte stored within an output-buffered packet switch is experiencing a delay greater than a given value.• Possibilities of reducing use of FDLs:

– Shared buffer (Multiple output channels use same FDL)– Multiple wavelengths in same fiber (used as FDL) => Wavelength converters.

94

TTM1:Approaches to Optical Internet Packet Switching

David K. Hunter and Ivan Andonovic

95 The design of Optical Packet Switches

• Three principal sub-blocks (Note: This is a slotted network):– Input interface: Alignment of packets i time. Why?– Switching core: Transports packets to the correct output port– Buffering using Fiber Delay Lines (FDL’s) (connected to switching matrix)– Output interface: Header insertion

96

Wavelength in Contention Resolution

• Broadcast and Select Switch (KEOPS)– Wavelength encoder. N wavelength converters, one for each input. Encoding each packet

on a fixed wavelength with a unique wavelength for each input.

– Buffer and broadcast section. Number of FDLs and a space switch stage. Electronically controlled selection (full signal?).

– Wavelength selector block. N demultiplexers, followed by electronically controlled selection.

• All packets available at all outputs => support multicast

97

Highlights you MUST know

98

Propagation through fibre

• Lightpulses are reflected into the core when hitting the cladding => approximately zero loss

Andreas Kimsås, Optiske Nett

99

Erbium Doped Fiber Amplifier (EDFA)• Widely deployed in optical networks

100

Chromatic dispersion

Figure: S. Bigo, Alcatel: Talk at Norwegian electro-optics meeting2004

101

Optical add/drop• Filter out a wavelength or a set of wavelengths. • Does not employ non-linear effects• Add a wavelength or set of wavelengths. • May be reconfigurable (ROADM)

– Select which wavelength to drop and add on a fibre

– Not as configurable as an optical cross connect (full cross-connection between several fibres)

1,2 1,2

11

Drop Add

102

4 different architectures1a) Fixed patch panel between WDM systems with transponders.

1b) Electrical switch fabric between WDM systems with transponders.

1c) Transparent switch between WDM systems with transponders, complemented by a OEO switch for drop traffic.

1d) Transparent switch on a transparent network. The signal stays optical until it exits the network.

103

•Consentrator => less fibers, needs power

•Many Fibers => no external power is needed

•Passive =>Higher power loss Do not need power

Fiber to the Home (FttH) variants

104

Scalability through VLAN hierarchy(4)

105

106

The design of Optical Packet Switches

• Three principal sub-blocks (Note: This is a slotted network):– Input interface: Alignment of packets i time. Why?– Switching core: Transports packets to the correct output port– Buffering using Fiber Delay Lines (FDL’s) (connected to switching matrix)– Output interface: Header insertion