WDM in backbone Péter Barta Alcatel-Lucent 10. October 2012
AGENDA 1. ROADM solutions 2. 40G, 100G, 400G 2
1. ROADM solutions 3
Ch 1-8 Ch 9-16 Ch 25-32 Ch 17-24 ROADM solutions What to achieve? Typical 1st gen (fix multiplexer architecture) WDM network: HUB OADM Ch 1-8 Physical WDM Ring Ch 25-32 OADM Ch 9-16 OADM Ch 17-24 OADM 4
ROADM solutions What to achieve? The targeted ROADM DWDM node provides: 1. Optical pass-trough and add/drop switching possibility for any single channel 2. Multidegree configuration in order to interconnect multiply rings and to offer the possibility to deploy meshed optical networks 3. Directionless arichitecture to allow to connect any client to any direction 4. Tunable multiplexer architecture to make short service provisioning times possible and to ensure restoration mechanisms to find spare routes if capcities are just available Just as in any state of the art network technology in the electrical domain: any port to any direction Line Line OT Line 5 Clients
Key element: WSS WSS does Mux/demuxing switching (between directions) spectrum alignment WSS offers (why WSS): mutidegree colorless all these in the optical domain (bit rate, protocol independent) 6
RX vs. TX WSS RX WSS Colorless ports by default Less optimal for long reach Efficient for Metro TX WSS Colorless ports not provided by default Optimized for long reach Efficient for Long Haul 7
RX WSS architectures 4-degree T/ROADM AMP IN Filter (East) THRU Filter (West) AMP OUT WDM IN CWR8 CWR8 WDM IN AMP OUT BB port 6 colorless ports BB port 6 colorless ports AMP IN FMUX Xpdr FMUX Xpdr Xpdr Xpdr Xpdr Xpdr FMUX Xpdr FMUX Xpdr WDM IN AMP OUT BB port 6 colorless ports BB 6 colorless ports port CWR8 CWR8 AMP IN WDM IN AMP IN THRU Filter (North) Filter (South) AMP OUT 8
Add/Drop channels WR8 RX WSS architectures Directionless/Colored R-OADM archistructure Local side Interconnectons Line Side TRBD TRBD X 6 5 4 3 2 1 OMDX WR8 WR8 WR8 WR8 Direction 1 Direction 2 Direction 3 Direction 4 9
TX WSS based architectures Directionless/Colorless (limited) TOADM Local A/D Block FMUX TX WSS MESH4 TX WSS MESH4 Mesh/Thru Connections TX WSS MESH4 TX WSS MESH4 TX WSS TX WSS AnyDir A/D Block 8 OTs RX WSS.. Multiple A/D block istances scaling up 8 OTs RX WSS
ROADM solutions What to achieve? 1. Optical pass-trough and add/drop switching possibility for any single channel 2. Multidegree configuration in order to interconnect multiply rings and to offer the possibility to deploy meshed optical networks 3. Directionless arichitecture to allow to connect any client to any direction 4. Tunable multiplexer architecture to make short service provisioning times possible and to ensure restoration mechanisms to find spare routes if capcities are just available 5. Changing the WDM channel ( recoloring ) in optical domain -> not achieved X Just as in any state of the art network technology in the electrical domain: any port to any direction 11 Lin e Lin e OT Clients Lin e
Functionality benchmark Meshed topology (multidirectional nodes) End to end mangement Free channel selection (colorless) Free direction selection (directionless) 1st gen WDM New gen WDM Technologies of the electrical domain (SDH, Ethernet, IP) No Yes Yes Partially Yes Yes No Yes * Yes No Yes * Yes * depending on the node construction 12
FLEXIBLE GRID Flexible Bandwidth the bandwidth of the WDM channel may vary (rate, modulation format, etc.) Optical switch fabric (e.g. WSS) with fine granularity (e.g. 3 or 6 or 12 GHz) without filtering between the adjacent grid elements -> channel bandwidth according to the actual need Fix Grid Flexible Grid
1/b Meshed operation Wavelength tracker 14
MU X Wavelength Tracker Operation Internal or Alien Integrated evoa s for source power management Sub-carrier modulation Payload (noise) WT Encode DSP Low Freq signal time Each channel is encoded with a unique WaveKey pair that allows the channel to be identified and its power monitored The assignment of WaveKeys is managed by the NEs, which maintain a database of the WaveKeys used in the network. At each detection point, the WaveKeys are detected and their power measured. DSP WT Decode AMP Uniquely identifies each service/wavelength in the network frequency Integrated per-channel evoas serve multiple functions Automatic optical power adjustment and unique wavelength coding Encode once, decode many times for fault location capability Provides a transponderless demarcation point Provides intra-node optical performance monitoring (OPM) at all line cards Decode based on DSP, correlation and orthogonal coding
Wavelength Tracker Encode/Decode Points WDM IN AMP IN THRU Filter (East) Filter (West) AMP OUT AMP OUT A A AMP IN WDM IN ADD/DROP = Wavelength Tracker Insertion Point = Wavelength Tracker Detection Point Transponder Each channel is encoded with a unique WaveKey pair that allows the channel to be identified and its power monitored. WaveKeys are encoded onto the channel at each Transponder and act as Optical J0 Trace Identifiers. The assignment of WaveKeys is managed by the NEs, which maintain a database of the WaveKeys used in the network. Wavelength Tracker WaveKeys are detected on the following cards: Amplifiers Wavelength Routers At each detection point, the WaveKeys are detected and their power measured. 16 Transponder
Alcatel-Lucent solution: Wavelength Tracker Features: Wavelength path trace Fault sectionalization & isolation Remote optical power control Threshold alarming Automated fault correlation Optical Fiber View Management of all wavelength on a selected fiber Optical Path View Management of all points along one wavelength Service-aware management of the optical layer
2. 40G, 100G, 400G Moving foreward in system capacity 18
Increasing the system capacity Ways of increasing the total capacity of a WDM system: increasing the number of channels denser grid broader used band increasing the data rate per channel using other physical properties like polarisation 19
Increasing the number of the channels Current status most commonly used band: C (~1530-~1565nm) -> effectively amplified by EDFA todays number of channels in C band: 40 (100GHz/0,8nm) 80 (50GHz/0,4nm) 10G/channel Increasing the number of channels use of broader band (e.g. introducing of band L) -> rather complex (and more expensive) amplifiers, concerns in case of a migration, residual CD issues denser grid in band C: less than 50GHz grid (e.g. 25GHz in case of 160 channels -> doubles only ones but blocks the way to higher bitrates (channel bandwidth) 20
Increasing the data rate per channel Needs rather complex modulation formats those of used up to 10G Steps until now: 2,5G: simplest NRZ/RZ OOK colored signals w/o any enchancements 10G: still OOK but use of FEC to ensure similar performances (reach) as that of 2,5G Interoperability issues between different modulation formats and bit rates channel type dependent reach: rather complex system designs By the time this way appears the most benefiting for system capacity upgrade. E.g. system capacities: with 10G channels: 80x 10G: 800Gb/s with 100G channels: 80x 100G: 8Tb/s 21
Transmission challenges beyond 10G Transmission beyond 10G is in general much more sensitive than 10G to the optical transmission impairments, e.g. 40G: 4 times more sensitive to optical noise (OSNR) 4 times more sensitive to fiber polarization-mode dispersion (PMD) 16 times more sensitive to fiber chromatic dispersion (CD) more sensitive to single-channel ( intra-channel ) nonlinear effects So, while at 10G the plain NRZ modulation format is sufficient to cover most LH/ULH applications, at 40G alternative modulation formats are needed to overcome these limitations Increasing transmission challenges requiring a new modulation format 22
DPSK at 40G new modulation format is need to overcome the impairments phase modulation instead of OOK less complex implementation: differencial PSK sufficient for 40G however can t catch 10G performance ~3dB improvment in OSNR tolerance but extended CD and PMD sensitivity remains direct detection with delay demodulation Adjacent bit is phase reference information in phase difference 23
Coherent detction for better tolerance Local oscillator is mixed with the incoming signal architecture to detect any level of phase modulations 24
Sampling Scope CD comp. Digital Clock Recovery Frequency and Carrier Phase recovery Symbol identification BER & Q² Re-sampling CD comp. Polarization Demultiplexing and Equalization Frequency and Carrier Phase recovery Symbol identification BER & Q² Digital signal processing PD1 ADC PD2 PD3 PD3 ADC ADC ADC DSP DSP j j 25
Rules of coherent receiver and DSP Coherent detection features: ability to detect absolute phase value separation of polarisations (basis for polarisation multiplexing) better sensitivity (detectors operate beside more optimal optical levels) DSP features: removing (compensating) linear impairments (CD, PMD) catching with the polarisation unstability -> enabling the polarisation multiplexing measure of the CD and PMD: supporting operation eliminating DCUs: in coherent only networks, less latency w/o DCUs the high CD value along the fiber mitigates non-linear effects -> extends the reach 26
Coherent detection in practice 40G: used to catch up with 10G reach 100G: the only way of implementing it 400G: also based on coherent detection wit further extensions Modulation formats: 40G: PDM-BPSK, 20 GBaud symbol rate 100G: PDM-QPSK, 25 GBaud 400G: PDM-16QAM dual carrier, 25 GBaud per carrier Symbol rate keep it low to fit in 50GHz spacing (80ch in C band) PDM be clearly above 10G to minimize the impairments casued by 10G channels 27
Further improvements in coherent transmission Issues to solve: 100G can t catch up in reach with pure coherent detection & DSP 400G is even more difficult Alcatel-Lucent s answer: Photonic Service Engine chip 28
400G PHOTONIC SERVICE ENGINE Working elements New modulations Dynamic demodulation HD SD Coherent Rx Enhanced algorithms Advanced frequency & phase recovery Tx DSP Wave shaping Ultra-fast ADC/DAC Higher bit converter Faster sampling 29
SD-FEC Not only a firm decesion is made if the bit is 1 or 0 as in case of classical HD-FEC but a probability is also provided needs more overhead 30
PSE applications universal tool for different cases can be used for 40/100/400G transmission in case of 40/100G it extends the reach in case of 100G the multiplier is >1,5x, needed to reach 10G / 40G coherent performance in case of 400G the enabler of implementation PSE Year 2010 2012 Speed 100 Gb/s 400 Gb/s Line rates 40G, 100G 40G, 100G, 400G Capacity 8.8T >23T Reach 2,000 Km > 3,000 Km Power/Gb* 6500 mw 4250 mw 31
2/b Nonlinearities 32
Power per channel [a.u.] Effect of nonlinearity Resilience to OSNR and fiber nonlinearity are key factors limiting the attainable transmission reach 1 db less resilience in nonlinearity is as limiting as 1 db less resilience to OSNR 20 15 10 5 0-5 Maximum power per channel to limit the nonlinear effects Attainable reach -10-15 -20 Minimum power per channel to guarantee the required OSNR 1 3 5 7 9 11 13 15 17 19 21 23 25 Number of spans 33
XPM 34
Penalty by XPM 35