1. OpenOLS & OpenDevice Overview
Stephane St-Laurent System Architecture
Dec 3, 2018

2. Agenda
Telecom Infra Project Overview
OpenDevice Model
OpenOLS Model

3. Open Optical Packet Transport (OOPT)
TIP OOPT WG Evolution
“We are creating a new approach
to building and deploying
telecom network infrastructure.”
Access
Backhaul
Core & Mgmt
TCLs
TIP Community Labs
TEACs
TIP Ecosystem Acceleration Centers
OOPT E2E Testbed
Open Optical Packet Transport (OOPT)
mmWave
OLS
CANDI
New
(Converged Architectures for Network Disaggregation & Integration)
C-API
Controls, Info Models and APIs (CIMA?)
Voyager
Cassini
DNP Platforms
DTC
Physical Simulation Environment (PSE)
Disaggregated Network Platforms (DNP)
New
Disaggregated Cell Site Gateways (DCSG)
New-ish
Odyssey

4. CIMA Mission Statement & Approach
Drive practical, multi-vendor implementations supporting operator-driven open optical use cases.
Approach:
Participate in and align closely to use case working group
Define a recommended set of control architectures, information models and APIs to enable focused implementation efforts and progress toward real-world deployment
Leverage best existing ideas from the industry, including ONF, OIF, MEF, Open ROADM, OpenConfig, etc.
Select a preferred subset for implementation focus
Develop contributions to enhance existing models/APIs, as needed to enable practical implementations
Focus on near-term feasibility for simplest, base use cases AND allow for increasingly complex and sophisticated use cases (multi-domain, multi-layer, multi-channel, E2E automation of planning, power control, restoration, etc.) 
Drive open multi-vendor interoperability demonstrations aligned to the target use cases and recommendations
Develop TIP OOPT testbed
Drive toward concrete, incremental milestones and expand in phased approach

5. Industry Progress on Info Models & APIs
(Objective)
OpenOLS,
OpenDevice
Alignment
CIM
TAPI 2.0, 2.1, …
TAPI 2.0 Interop Demo
~Alignment
ODTN
1.0, 1.5, 2.0…
Transponder Control

6. Multi-vendor OLS POC – TIP Summit 2017
Architecture can be adapted to all expected use cases

7. Multi-vendor OLS POC – TIP Summit 2017 Global View | ROADM Degrees + Line Amplifiers
Open OLS
Yang Model Definition
Open OLS
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
Very short: This is the Topology View that expose the OpenOLS model
Abstraction Layer (OpenOLS)
Nodal View: ROADM Degree and Optical Transmission Line ILA
Supports connection of a media channel (single or multi-carrier)

8. Objective I Define a model that is:
Based on standard model or/and use standard concept
Is generic enough to be widely deployed across multiple product (Integrator/Component)
Support Centralized/Distributed/Hierarchical Power Control in Controller/Manual mode
Implementation that support Access/Metro/Regional/Long-Haul/Subsea
Can be push into open source as model (and may be a reference to improve other standard model)
Ex: ONF TAPI, ONF ODTN, ONF CIM, MEF, ITU-T (through ONF), OpenROADM. OpenConfig
Agnostic to implementation/composition
Can be automated for instantiation through schema, scripting, etc

9. Objective II
Support those 3 concepts

10. Objective III System Knowledge No System Knowledge OpenOLS (YANG)
Abstract layer that receive intent but understand the recipient of the intent
Composition
Describe that F = E(F1)+ E(F2)
No System Knowledge
OpenDevice (YANG)
Low level API that exposes basic optical function and their specifications
Schema:
This is a declarative model (description of what is offered)

11. Where it come from What it is Why we want standardization
OpenDevice
Where it come from
What it is
Why we want standardization

12. OpenDevice Root Annex 4 of CIM model (ONF) by Nigel Davis (CIENA)
OpenConfig
Configuration & State separation
YANG base
Simplicity
Lumentum Model
Practical
Optical Component based
Based on Optical Function
Card Composition (Local scope, not system scope)
OpenROADM
Device model
Composition
Power Control model (Metro based)
Several other Component Vendor model
Finisar (WSS), Oplink Discussion, II-VI

13. OpenDevice | Equipment Optical Function
Construct Schema
ROADM-Degree
opendevice-schema
Includes: equipment, ports, functions, inter-connections
System IN
Line OUT
WSS Mux
Primitives
opendevice-wss
System OUT
Line IN
WSS DeMux
opendevice-edfa
opendevice-voa
Line-Amp
opendevice-power-monitor
opendevice-channel-monitor
Line IN
Line OUT
opendevice-equipment
opendevice-equipment:port
Line OUT
Line IN
Includes: Control, Operational State, & Specification

14. Basic Forwarding Construct
The filter models the ability to allow only those photons that are within in a defined portion of spectrum to be passed. The filter is described as a media channel and is represented by an FC.
The portion of the spectrum is called a frequency slot and is described by centre frequency and width. Frequency slot is an administrative concept and is conceptually square. The actual pass-band of the filter is not square. The frequency slot and pass band relationship is challenging and not covered here.
A single port of a filter can support more than one media channels (see later).
As the filter is represented by an FC the characteristics are expressed in an FcSpec (see TR-512.7).
The model (and symbol set) allows for a tuneable attenuator.

15. Coupler/Splitter
The Coupler/Splitter provides a set of atomic media channels between one (common) port and two or more other (branch) ports.
All of these atomic media channels have the same frequency slot. In the root to leaf direction the "splitter" attenuates the signal, in the leaf to root direction the "coupler" attenuate the signal.
Coupler and Splitter with Filter could be combined

16. OpenDevice: Specialization
Expose:
Configurable Control Function
Operational State
Optical Specification
Optical Specification example
EDFA
Operating Spectrum
Gain Masks
Tilt Masks
Noise Figures
WSS
Attenuation Range
Min/Max Channel Width
Timeslot Granularity

17. Sensor (PD and Channel Monitor)
No loss
Bounded in Frequency
FCM:
2 modes
channel-monitor
Status only
List of frequency slot with power
flex-channel-monitor
Configurable (SMC that contains NMC)
Use lower and upper frequency
Status could provide detected signal

18. EDFA Bounded in frequency Optical Control model (Gain or Power)
Enabled/Disabled
Target Gain / Target Tilt
Target Power
Gain Range Selection

19. Wavelength Selective Switch (WSS)
Bounded in frequency
List of Connection (SMC passband) with source/destination port
Connection could be blocked (True/False)
Define by lower/upper frequency
Control mode selection
Direct: OpenLoop
Power: Adjust to the target power value
Attenuation: Adjust to the target attenuation
Frequency Slot:
Relative attenuation in the passband that is divided by the value associated to the device frequency slot granularity

20. VOA
Control:
Direct
Attenuation
Power
Blocked (True/False)

21. RAMAN (backard) Apply the gain to the ingress fiber
Bounded in frequency (C, L or C+Lband)
Optical Control model (Direct only)
Enabled/Disabled
Target Gain / Target Tilt
Need fiber type
Point-Lost offset

22. OpenDevice: optical-control-mode
Forwarding Construct (FC) are defined by passband (lower-frequency, upper-frequency)
Forwarding Construct (FC) expose a transfer function that could be fixe or configurable, when configurable, it contain an optical-control-mode (direct-control, gain-attenuation-control, power- control)
In direct-control, there is no sensor required since it operate in open-loop
In gain-attenuation-control, there is an ingress and an egress sensor since the device operate in close-loop control. The control-loop could be enabled (ON) or disabled (OFF). The sensors are part of the device schema and, if required, are configured by the device itself.
In power-control, there is an egress sensor since the device operate in close-loop control. The control-loop could be enabled (ON) or disabled (OFF). The sensor is part of the device schema and, if required, is configured by the device itself.

23. OpenDevice Use Config Section
Use State Section ( to reflect the config section and to provide operational stte)
Include a reference to a specification
Include a Base and a System Section (Manual/Automatic control model)

24. Power Monitoring

25. OPM in ITU
Optical Power Measurement is define in ITU to monitor the total power associated to media channel, its mean that it is bound to select spectrum

26. OPM Location
Most implementation translate the location of the measurement point to a location at the demarcation point close to the pin (port)
This provides the best performance for EDFA gain and relate to the specification of the equipment
OpenROADM and Facebook TIP use that location for OPM

27. Base Model
Need to monitor the power for each of the media channel

28. Model 1 (not good)
Could be represented using tap coupler and pin monitor in the FC chain
The implementation normally assume that the pin monitor is compensated to the demarcation point but this could not be obvious to deduct
In this implementation, it is not the right spectrum that is monitored
Not good, measurement is bound to media channel spectrum

29. Model 2
In this implementation, the OPM reflect the power associated to the spectrum. It is the optical power seen in the media channel
Again, translation is required to present the information at the demarcation point

30. Proposal to define FC to represent PD Sensor
PD Sensors FC contain filtered Pin Monitor
The FC expose no loss from the input to the output with translation (Other FC will be included to represent effective lost where it should be present in the schema)
It contain list of OPM with spectrum definition
They could be place in the FC chain where we want to see the measurement
FC port of other FC can delegate their measurement to that PD Sensor

31. PD Sensor
The FC port were the media channel start delegate the measurement to the PD Sensor FC
The sensor is located to the wanted point of measurement
It present a Band Power measurement

32. Flex Channel Monitor Sensor
Possible Location for Channel Monitor
Only requirement from ITU is to detect if an NMC is present to feed the OTSiA OAM
For Control, it will be preferable to report SMC and NMC at the Demarcation point

33. Band Monitor (SMC and NMC)
The SMC/NMC power need to be consistent with their client server relation.
It could be translated to the demarcation point
Reported Power is spectrum bounded

34. OPM Monitor location

35. OPM
The TAPI stack expose CTP and TTP, those CEP present OPM package and the location could be referred into the Optical FC Schema (OpenDevice Schema in Facebook TIP)
The OPM stack maintain the OPM validity because of the client-server relationship

36. OpenDevice
Data Model (YANG)

37. opendevice-edfa (example)
List of edfa
Control: gain/power, tilt
Gain-range selection
Enabled
State
ingress-sensor
egress-sensor
psd-ase-actual-mc-band

38. opendevice-voa (example)
List of voas
Control
Direct, Attenuation, Power
Blocked
Ingress-sensor-ref
Egress-sensor-ref
State
ingress-sensor
egress-sensor

39. opendevice-wss (example)

40. OpenDevice Schema

41. OpenDevice Schema (The equipment detail)
Describe what you are !
This is an equipment
This is the list of my ports
This is the list of my basic function
This is how it is interconnected

42. OpenDevice Schema Each equipment has an OpenDevice schema
List of ports
List of links with reference to equipment/port or device/port
List of forwarding construct with fc-port
Forwarding construct are recursive following the same model as CIM
The forwarding construct have reference to OpenDevice model

43. CTRL: ROADM Degree and ILA

44. EDFA Example

45. OpenDevice Schema Model
YANG Data model to be provided later
We are currently editing it

46. Specification
Gain Mask

47. Gain mask Example of switched-gain EDFAs (low & high gain)
Simple gain mask defines valid operating regions:
Normal (controlled tilt)
Extended gain (uncontrolled tilt)
No additional mask features (e.g. drop down regions)
We can represent the gain mask using two complex polygons
Normal: ((x2,y2), (x3,y3), (x4,y4), (x5,y5))
Extended: ((x1,y1), (x2,y2), (x5,y5), (x6,y6))
Diagonal line represents the current gain setting
We can project the min/max points of this gain line onto the x-axis. This gives the allowed input power range for the current set gain
(x6, y6)
(x5, y5)
(x4, y4)
Normal gain
Extended gain
GEDFA = yi – xi
Output Power (dBm)
(x1, y1)
(x2, y2)
(x3, y3)
Input Power (dBm)

48. Gain mask validation: Simple example
Example: Normal gain region validation
We want to determine the valid range of input powers, for a given gain Gi: [Pin_min(Gi),Pin_max(Gi)]
From the gain mask shape, we can determine the equation of max/mib
Pin_min(Gi) = ((x3-x2)*(y2-Gi) - x2*(y3-y2))/((x3-x2)-(y3-y2))
Pin_max(Gi) = ((x4-x3)*(y3-Gi) - x3*(y4-y3))/((x4-x3)-(y4-y3))
Exceptions:
In cases where the lines collapse to a single point (e.g. x2 == x3 AND y2 == y3), the above equations blow up due to DIV by 0.
Need to force calculated power to:
Pin_min(Gi) = x2 (or =x3)
See next example for further details. Gi
GEDFA_normal_min
GEDFA_normal_max
Valid input
powers
Pin_max(Gi)
Pin_min(Gi)
(x5, y5)
(x4, y4)
Output Power (dBm)
(x2, y2)
(x3, y3)
Input Power (dBm)

49. Gain mask validation: complex example
This example is analogous to the FRM-20X MUX EDFA gain mask
Three regions:
Normal ((x2,y2), (x3,y3), (x4,y4), (x5,y5))
Extended gain ((x1,y1), (x2,y2), (x5,y5), (x6,y6))
Drop-down ((x7,y7), (x7,y7), (x2,y2), (x1,y1))
3 complex polygons can completely describe these regions for all supported gain values
4 points per polygon, two points for each gain setting
Note how the triangle is defined to handle the point
(x6, y6)
(x5, y5)
(x4, y4)
Output Power (dBm)
Extended gain
Normal gain
Extended gain
(x1, y1)
(x2, y2)
(x3, y3)
Drop-down
Input Power (dBm)
(x7, y7)

50. Gain mask validation: complex example
Normal mask: ((x2,y2), (x3,y3), (x4,y4), (x5,y5)) Pin_min(Gi) = ((x3-x2)*(y2-Gi) - x2*(y3-y2))/((x3-x2)-(y3-y2)) Pin_max(Gi) = ((x5-x4)*(y4-Gi) - x4*(y5-y4))/((x5-x4)-(y5-y4))
Extended mask: ((x1,y1), (x2,y2), (x5,y5), (x6,y6)) Pin_min(Gi) = ((x2-x1)*(y1-Gi) – x1*(y2-y1))/((x2-x1)-(y2-y1)) Pin_max(Gi) = ((x6-x5)*(y5-Gi) – x5*(y6-y5))/((x6-x5)-(y6-y5))
Drop-down mask: ((x7,y7), (x7,y7), (x2,y2), (x1,y1)) Pin_min(Gi) = x7 Pin_max(Gi) = ((x2-x1)*(y1-Gi) – x1*(y2-y1))/((x2-x1)-(y2-y1)) Gi
Pin_max(Gi)
Pin_min(Gi) Gi
Pin_max(Gi)
Pin_min(Gi)
(x6, y6)
(x5, y5)
(x4, y4)
Output Power (dBm)
Extended gain
Normal gain
Extended gain Gi
Pin_max(Gi)
Pin_min(Gi)
(x1, y1)
(x2, y2)
(x3, y3)
Drop-down
Input Power (dBm)
(x7, y7)

51. Example gain mask features
y = x + GEDFA
Some EDFA operate also with single extended gain drop-down
It is map as a single region where 2 of the 4 points are similar
In this example, there is 3 regions
Single extended gain drop-down

52. Specification
Tilt Mask Model

53. Tilt mask Mask#1 Mask#2 We can specify tilt with a tilt mask
Based on the mask’s complexity, we break it down into two quadrilaterals (four-sided polygons)
Projecting a valid Tilt range for gain setting Gi
Slicing the mask with respect to the gain axis
Mask1: ((x1,y1), (x2,y2), (x4,y4), (x5,y5))
Mask2: ((x2,y2), (x3,y3), (x3,y3), (x4,y4))
Allowed tilts in Mask1: Tiltmax = ((y5-y4)/(x5-x4))*(Gi-x4) + y4 Tiltmin = ((y2-y1)/(x2-x1))*(Gi-x1) + y1
Allowed tilts in Mask1: Tiltmax = ((y4-y3)/(x4-x3))*(Gi-x3) + y3 Tiltmin = ((y2-y1)/(x2-x1))*(Gi-x1) + y1
Tiltmax(Gi)
Tiltmin(Gi)
Tiltmax(Gi)
Tiltmin(Gi)
(x5, y5)
(x4, y4)
(x3, y3)
Mask#1
Mask#2
Tilt (dB)
(x1, y1)
(x2, y2)
EDFA Gain (dB)

54. Specification
Noise Figure Model

55. Gain vs. NF curves
Gain vs. NF curves can be used to estimate amplifier performance at the current operating condition (gain & tilt)
NF is tilt dependent
Options for modelling:
Highest level Use max NF spec across all tilts for a single curve
High level Use NF spec data points and curve fitting for different tilts
Lowest level Use mfg. test data stored in EEPROM + aging margin to estimate actual HW performance over life
Tilt1
(x1, y1)
Tilt2
NF (dB)
Tilt3
(x2, y2)
(x3, y3)
(x4, y4)
(x5, y5)
Gain (dB)

56. Specification
ASE Model

57. ASE Model Current model is to provide ASE at Run-time
EDFA can be query to retrieve ASE as a PSD value
Raman can be query to retrieve ASE as PSD values
Unit: mW/nm

58. Where it come from What it is Why we want standardization
OpenOLS
Where it come from
What it is
Why we want standardization

59. OpenOLS root
OpenOLS root are the CIM model after it went to refactoring to produce TAPI
Provides an abstraction model
Layer Agnostic
Support Model for different Service
Connectivity, Topology, OAM. Equipment (next release), Resource management (Talking about it)
Very Generic
You could build anything (like in CIM model)
Derive specialization from standard body (MEF, ONF, IEEE, ITU)
Use conversational interface (CRUD) (Config is through RPC, not set)
YANG model
Work with Restconf, Netconf and YANG
OpenROADM
Give the node some specialization
ROADM, DEGREE, SRG, ILA

60. Original Degree Base model use in Facebook POC Degree of a ROADM
Operation: Connection Config
Channel contains Carriers
Use PSD
Start, stop frequency
Composition

61. Multi-vendor OLS POC – TIP Summit 2017 Global View | ROADM Degrees + Line Amplifiers
Open OLS
Yang Model Definition
Open OLS
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
OpenDevice
Yang Model Definition
Very short: This is the Topology View that expose the OpenOLS model
Abstraction Layer (OpenOLS)
Nodal View: ROADM Degree and Optical Transmission Line ILA
Supports connection of a media channel (single or multi-carrier)

62. To This: ROADM Node
C&L Stack

63. Simple Connection (Single Carrier)
Operational View of a connection
Full abstraction model
A connection in a ROADM node

64. More Complex Connection (Multi-Single Carrier)
OTSiA
Multiple SMC with single NMC

65. More Complex Connection (Multi-Carrier)
OTSiA
SMC with multiple NMCs

66. OLS Controller (NE)
Use refactored TAPI (simplified since declarative connectivity-service, topology, connectivity, OAM)
Expose ROADM node with encap-topology
Or Directly Degree node for disaggregated NE
Contain Composition (Node that represent equipment)
Expose OpenDevice to provide optical power control

67. Photonic Media Definition
Media Channel: optical spectrum bandwidth between end points in the photonic layer
Network Medial Channel (NMC): continuous optical spectrum between end points in the photonic layer to represent the optical spectrum intended to carry a signal
Service Media Channel (SMC): continuous optical spectrum between end points in the photonic layer obtain through optical filter that serve NMC (optionally SMC)
Assembly: group of media channel (Network or Service Media Channel) managed as a single entity

68. SMC/NMC Service Media Channel (SMC) Network Media Channel (NMC)
Continuous spectrum bandwidth define by lower and upper frequency
Could be aligned on ITU grid (could be 6.25 GHz)
Include guardband (could be 6.25 GHz and it is technology dependent)
SMC are indivisible when using WSS technology (port to port where the NMC have to be routed together)
SMC could contain 0 to n NMC
SMC could be requested at a domain level to represent the contiguous spectrum offered by an optical domain
Network Media Channel (NMC)
Continuous spectrum bandwidth used to represent the signal component generated by an optical transmitter
Define by center frequency and width or lower frequency and upper frequency
Do no require to be aligned to ITU grid but are technology dependent (channel power monitor and transmitter capability)
Should not overlap guard-band of SMC

69. Service Media Channel
A service media channel (SMC) is define by the optical spectrum between its lower frequency and its upper frequency
It contain guard bands (read only), a lowerGuardBand and an upperGuardBand, that represent a filter specification (will be define later)
The constraint apply to the possible value for the lower and upper frequency

70. Network Media Channel
A network media channel (associated to the signal spectrum) is define by the optical spectrum between its lowerFrequency and its upperFrequency
It also provide a centerFrequency and its nmcSpectrumBandwidth
The constraints apply to the centerFrequency value

71. NMC Power Measurement
The measurement of the power need to be define by a bandwidth, It need to correlate between transponder (OTSi) and Optical Line System (NMC)
The nmcSpectrumBandwidth is used to measure the power associated to the signal between the lower and upper frequency

72. NMC detail
The transmitter (OTSi) own the accuracy of the signal center frequency so it could determine the required nmcSpectrumBandwidth (MHz) associated to the Network Media Channel (NMC)

73. Multi Carrier Detail
When multiple NMC could be contain inside one SMC, a user should be able to pass information so the NMC could be adjacent, or with some spacing, to each others
Adding some spectrum spacing could improve OSNR of the signal
Note: This provides to the Path Computation Engine (PCE) the information needed to allocate the center frequency to the NMCs. The PCE must still maintain the constraint associated to the centerFrequency that is defined by the transponder (OTSi) specification

74. Parameter used for multi-carrier SMC
Plan of Record (POR)
Spacing between NMC could be passed using nmcAdditionalSpectrum or nonAdjacentNmcSpectrum. One parameter need to be agree on. The nonAdjacentNmcSpectrum give a better understanding to evaluate non linear impairment (NLI).

75. SMCs Constraint SMC spectrum do not overlap
The SMC frequency of 2 adjacent SMCs could be however equal
upperFrequency (uf) of SMC1 could be equal to the lowerFrequency (lf) of the adjacent SMC2

76. Media Channel Pool
A media channel pool define the spectrum that is supported by the pool
This is define as a list of MC pool specifications

77. Parameter for SMC Pool Specification
The Media Channel Pool provides the supportableSpectum [lf, uf] parameter. This is to expose what the node-edge- point (NEP) pool is capable.
Connection are between connection- end- point (CEP) and capability of the connection will be expose through constraint (gridType, granularity, etc)
SMC CEP could be define from the pool supportableSpectrum but cannot extend outside the pool spectrum limit. The frequency value associated to the SMC CEP could be however equal to the limit of the pool spectrum.

78. Upper and Lower Frequency constraint related to SMC
A forwarding domain can expose, through topology, the constrain associated to the SMC granularity for the lower and upper frequency
The 2 examples on the left show case of granularity of 12.5 or 25 GHz

79. Connectivity Service: SMC/NMC
OpenOLS use a declarative model for the connectivity service. It is not an “Intent” like in TAPI connectivity service. The connection has already been resolve by the PCE of the SDN controller or the PCE of GMPLS
The number of SMC/NMC are known, the passband lower-frequency and upper- frequency are known, the references PHYSICAL sip and SMCA sip are known
SMC only connectivity service request are used to preconfigure passband that will be used to contain SMC

80. Connectivity Services: SMCA/NMCA
Facebook TIP YANG Model Demonstrated
Connection are SMCA
SMCA contains a list of SMC
SMC contain a list of NMC
Updated for TAPI (YANG)
SMCA contains a list of NMCA pool (only 1 needed)
NMCA pool contains the list of NMCA
NMCA contains a list of NMC

81. Side by Side: Config and Operational

82. C and C+L (SMCA/NMCA/NMC not shown)

83. TIP Model vs New Model

84. TIP vs ODTN

85. Including Orchestrator and Multi-Domain
Orchestrator Resolve Resource in the multi-domain
Each SDN (OLS controller) manage their own domain
L0 Power Controller manage interaction between domain