Virtual Prototyping of High-Performance Optical Networks for Advanced Avionics Systems

1
Virtual Prototyping ofHigh-Performance Optical
Networks for Advanced Avionics Systems

• Dr. Alan D. George
• Ian Troxel, Jeong-Hae Han, Nang Dilakanont
• Jeremy Wills, Todd McCaskey
• High-performance Computing and Simulation (HCS)
  Research Laboratory
• Department of Electrical and Computer Engineering
• University of Florida

2
Outline

• Introduction
• Tool Evaluations
• MLDesigner Overview
• Preliminary Modeling
• Potential Applications
• Conclusions

3
Introduction

• Optical Networks for Advanced Avionics Systems
• Driving force key applications for future
  avionics systems (cockpit, cabin)
• Pushing performance and reliability requirements
  to ever-increasing levels
• Integrated networking infrastructure
• Higher bandwidth, deterministic performance,
  fault tolerance, lower cost
• Solutions will come from new and emerging
  technologies in high-speed optical networks
• e.g. WDM, 10 Gigabit Ethernet, etc.

4
Introduction

• Optical networking technologies rapidly
  progressing, complex
• Optimal means to select, adapt, combine, and
  deploy for advanced avionics systems at lowest
  cost is difficult to ascertain or anticipate
• Research required to investigate strengths,
  weaknesses, and tradeoffs
• Complexity limits usefulness of analytical
  methods
• Cost and time constraints limit usefulness of
  experiment methods
• For both existing and especially emerging
  concepts and technologies
• Primary approach will be computer-based
  simulation
• Supported by selected analytical and experimental
  methods
• Our project focuses on development and
  exploitation of such tools
• Rapid virtual prototyping of high-performance
  optical networks
• Networks as integral part of advanced avionics
  systems

5
Introduction

• Four phases of FY03 project
• Evaluation of simulation tools (completed)
• Construction of component models (underway)
• Construction of system models (future)
• Simulation experiments and analysis (future)

6
Introduction

• Team Modeling Experience
• Network modeling
• SCI and SCI/RT networks (BONeS, UltraSAN)
• Myrinet network (BONeS)
• Fibre Channel network (BONeS)
• Architecture and systems modeling
• RISC (BONeS, MLDesigner)
• CMP (C, extended SimpleScalar)
• SMP (BONeS)
• Reconfigurable network processor (BONeS)
• HWIL and SWIL simulation (BONeS)
• New efforts underway
• Optical avionics networks (MLDesigner) our
  primary focus
• FPGA-based RC architectures and systems
  (MLDesigner)
• End-to-end performance modeling for data grids
  (MLDesigner)
• Dependability modeling for mission assurance
  (MLDesigner)

7
Introduction
Facilities for Simulative and Experimental
Research

• Cluster-based HPC
• Dell and custom boxes
• 376 Pentium-compatible CPUs
• 240 nodes 300 GFLOPS peak
• 50 GB main memory
• 3.1 TB storage
• PCI64/66 support (i.e. 4?PCI)
• Networking testbeds
• 5.3 Gb/s SCI
• 2.5 Gb/s InfiniBand
• 1.28 Gb/s Myrinet
• 1.25 Gb/s Giganet cLAN
• 1.0 Gb/s Gigabit Ethernet
• 155 Mb/s ATM

8
Tool Evaluations

• Initial Goals for Optical Networking Tool
• Model networking issues (data link, network
  layers, etc.) while maintaining a realistic
  representation of optical physical layer
• Desire for library of pre-built models
• Stability and maturity
• Responsive technical support
• Reasonable cost

9
Tool Evaluations
Networking Networking
MERLiN GLASS
NIST Simulators ARTHUR
ONSS cnet
Real / NeST QualNet
OptSim Transmission Maker
ns-2 WDM Guru

• Divergent Roads
• Networking tools
• Protocols and topology focus
• Typically open source
• Physical layer abstracted
• Optical tools
• Optics focus
• Typically expensive
• No networking protocols
• Others
• Various strengths and weaknesses

Optics Optics
Artifex LinkSim
Transport Maker OptiSystem
PHOTOSS PHOTOSS
Others Others
HyPerformix Workbench Simulink / Matlab
MLDesigner BONeS
OPNET Modeler OPNET Modeler
10
Tool Evaluations
11
Tool Evaluations

• MLDesigner selected as best all-around tool
• Flexible
• Models fully extendible and user-definable
• Supports different modeling domains with high
  fidelity
• HCS lab is currently building an optical
  networking library
• Industry acceptance and technology support
• Aerospace Corp, Agere, Apple, Astrium, Ericsson,
  ifEN (Germany), Infineon, KPN (Netherlands),
  Lockheed Martin, Motorola, Philips Research,
  Rockwell Collins, Siemens, etc.
• Cost effective
• 7-9K annual corporate license (per seat)
• Free for academic institutions
• Knowledge base
• Builds upon BONeS (previous lab experience)

12
MLDesigner Overview

• MLDesigner is an integrated platform for modeling
  and analyzing the architecture, functionality and
  performance of system designs.
• Multi-domain simulator for design and analysis of
  a broad range of applications.
• Interfacing to SatLab and MATLAB/Simulink extends
  MLDesigner capabilities.
• A system model is constructed through the
  graphical editor or the PTcl command language.
• MLDesigner employs a hierarchical block-level
  design.
• BONeS and COSSAP models can be imported.

13
MLDesigner Overview
Domain Description
Discrete Event (DE) An event-driven model of computation. Events are processed in chronological order.
Dynamic Data Flow (DDF) Dynamic scheduling. Supports conditionals, data-dependent iterations, and true recursions. Good for signal processing applications.
Synchronous Data Flow (SDF) Static scheduling. Flow is completely predictable at compile time. Good for synchronous signal processing systems with multi-rate applications.
Boolean Data Flow (BDF) Compile-time scheduling. Trying to achieve efficiency of SDF but generality of DDF. Execution of a task is annotated with a Boolean condition.
Continuous Time/Discrete Event (CT/DE) Combined continuous-time and discrete-event model of computation.
Finite State Machine (FSM) Supports finite state machine elements (i.e. events, states, transitions, actions, arguments, histories).
High Order Function (HOF) Implements behavior of functions that may take a function as an argument and return a function.
14
MLDesigner Overview

• Graphical editor, including parameter and DS
  editor.
• PTcl command environment to define complex system
  that is difficult to define graphically.
• Multi-domain simulators including debugging
  animators.
• Module functionality can be specified by
  hierarchical block diagram, finite state machine,
  user primitive (C/C), or PTcl module
  definition.
• Simulation results can be viewed through
  animation during simulation and/or by
  post-processing graphical plots.

15
Preliminary Modeling

• Current Modeling Status
• Component modeling began on 1/10/03
• Optical data structures defined
• Many internal components built and verified
• Building blocks for modules (e.g. power loss, BER
  injection, etc.)
• Several modules built and verified
• Laser, Tunable Receiver, Transmitter, Fiber,
  Power Amplifier, 1x2 Splitter, 2x1 Coupler
• Others in progress
• MUX, DEMUX, OADM, Star Coupler
• Small WDM system model built with 3 simple tests
  performed

16
Preliminary Modeling
Single Optical Link DS (Single_OL_DS) Single Optical Link DS (Single_OL_DS) Single Optical Link DS (Single_OL_DS) Single Optical Link DS (Single_OL_DS)
Field Type Init. Value Comment
Higher-layer Data Int -1 Bits to be transferred over the link.
Number of Bits Int -1 Number of bits in Higher-layer Data.
Wavelength Float -1.0 Transmission wavelength (nm).
Current Power Level Float 100.0 Current signal power level (dB).
Creation Time Float -1.0 Time the DS was created.
ID Number Int -1 ID number for test purposes.
WDM Optical Link DS (WDM_OL_DS) WDM Optical Link DS (WDM_OL_DS) WDM Optical Link DS (WDM_OL_DS) WDM Optical Link DS (WDM_OL_DS)
Field Type Init. Value Comment
Number Active Waves Int -1 Number of waves currently in the WDM.
Active Waves Float Vector -1,-1, A list of the wavelengths currently in the WDM.
Active Wave DSs Single_OL_DS Vector Initialized Single_OL_DS The DSs for the active waves.
Creation Time Float -1.0 Time the DS was created.
ID Number Int -1 ID number for test purposes.
17
Preliminary Modeling
Internal Components Internal Components Internal Components
Name Description Status
Cross Wave Interference Models the influence a given wavelength has the on power and bit errors of another wavelength. In Progress
Extract Fields Single_OL_DS Displays all the inner fields of a Single_OL_DS. Verified
Extract Fields WDM_OL_DS Displays all the inner fields of a WDM_OL_DS. Verified
Graph Single_OL_DS Produces a plot of the inner fields of a Single_OL_DS Verified
Graph WDM_OL_DS Produces a plot of the inner fields of a WDM_OL_DS Verified
Optical Power Gain Models optical power gain applied to a given wavelength Verified
Optical Power Loss Models optical power loss applied to a given wavelength. Typically based on fiber length or coupling loss. Verified
BER Injection Models device bit error rate injection. In Progress
Place Single in WDM Determines if a given wavelength can be added to a WDM signal. An error is generated if that wavelength already exists or if the WDM signal is at a predefined maximum. Verified
Remove Single from WDM Determines if a given wavelength is in a WDM signal. An error is generated if the wave is not present. Verified
Single_OL_DS Create Creates an un-initialized Single_OL_DS Verified
WDM_DS Create Creates an un-initialized WDM_OL_DS Verified
Modules Modules
Name Status
1x2 Splitter Verified
2x1 Coupler Verified
4x1 MUX In Progress
1x4 DMux In Progress
Laser Verified
Tunable Receiver Verified
Transmitter Verified
Fiber Verified
Power Amplifier Verified
OA/DM In Progress
Star Coupler In Progress
18
Preliminary Modeling
Optical Fiber
Length (Km) ? Loss Per Unit Length (dB/Km)
Length (Km) ? Delay Per Unit Length (us/Km)
Optical Out
Interference Loss
Bit-Error Injection
Optical In
Optical Amplifier
Power Gain Per Wavelength (dB)
Coupling Loss (dB)
Coupling Loss (dB)
Optical In
Optical Out
Bit-Error Injection
Interference Loss
19
Preliminary Modeling
Optical Transmitter
Generate New Single_OL_DS
Packet ID Value
Coupling Loss (dB)
Insert The Data
Data Value In
Power Variation Function
Fixed Output Power
Laser Source
Power Out
Any Event In
Wavelength Out
Fixed Output Wavelength
20
Preliminary Modeling
Change Receiver Wavelength (nm)
Tunable Optical Receiver
Initialize Receiver Wavelength (nm)
Select Wavelength (nm) from Input
Data Out
Data In
Coupling Loss (dB)
Check Power Threshold (dBm)
21
Preliminary Modeling
Try to Place Channel 1 in New WDM Signal
Try to Place Channel 2 in New WDM Signal
2x1 Coupler
Channel 1 In
Increment WDM ID
WDM Out
Channel 2 In
If WDM Signal Empty -- Discard
Coupling Loss (dB)
Check for Wave Interference
Coupling Loss (dB)
22
Preliminary Modeling
1x2 Splitter
Coupling Loss (dB)
Channel 1 Out
OL In
Channel 2 Out
Coupling Loss (dB)
23
Preliminary Modeling
Simple Test System

• Tests Performed
• End-to-End Delay
• Receiver Power Threshold
• Wavelength Selection

• Components Included
• Optical Transmitter
• Optical Fiber
• Optical Power Amplifier
• 2x1 Splitter
• 1x2 Coupler
• Tunable Receiver

24
Preliminary Modeling

• Delay Test Setup
• Two Sources
• Source One
• Wavelength 1530 nm
• Data values step by 1 from 0
• Output power level at 7 dBm
• Source Two
• Wavelength 1560 nm
• Data values step by 1 from 10
• Output power level at 7 dBm
• Optical Fibers
• Length 40 Km
• Power loss per Km .25dB/Km
• Delay per Km 3.3 us/Km (velocity c)
• Power Amplifiers
• Power gain 10 dB

• 2x1 Coupler
• Coupling loss .17 dB
• 1x2 Splitter
• Coupling loss .17 dB
• Two Tunable Receivers
• Receiver One
• Wavelength 1530 nm
• Input coupling loss .17 dB
• Power threshold level 17 dBm
• Receiver Two
• Wavelength 1560 nm
• Input coupling loss .17 dB
• Power threshold level 17 dBm

25
Preliminary Modeling
ID
ID
Source
Source
Receiver
Receiver
Cycles
Cycles

• Delay Test Results
• Packets for each source / receiver pair are
  delayed by appropriate amount

26
Preliminary Modeling

• Power Threshold Test Setup
• Two Sources
• Source One
• Wavelength 1530 nm
• Data values step by 1 from 0
• Output power level varies from -7 dBm
• to 17 dBm and back in steps of 1 dBm
• Source Two
• Wavelength 1560 nm
• Data values step by 1 from 10
• Output power level varies from -7 dBm
• to 17 dBm and back in steps of 1 dBm
• Optical Fibers
• Length 35 Km
• Power loss per Km .25dB/Km
• Delay per Km 3.3 us/Km (velocity c)
• Power Amplifiers
• Power gain 10 dB

• 2x1 Coupler
• Coupling loss .17 dB
• 1x2 Splitter
• Coupling loss .17 dB
• Two Tunable Receivers
• Receiver One
• Wavelength 1530 nm
• Input coupling loss .17 dB
• Power threshold level -17 dBm
• Receiver Two
• Wavelength 1560 nm
• Input coupling loss .17 dB
• Power threshold level -17 dBm

27
Preliminary Modeling
dBm
dBm
Source
Receiver
Source
Receiver
Cycles
Cycles

• Power Threshold Test Results
• Packets for each source / receiver pair are only
  received when power level is at or above receiver
  threshold of -17 dBm

28
Preliminary Modeling

• Wavelength Select Test Setup
• Two Sources
• Source One
• Wavelength 1530 nm
• Data values step by 1 from 0
• Output power level at 7 dBm
• Source Two
• Wavelength 1560 nm
• Data values step by 1 from 10
• Output power level at 7 dBm
• Optical Fibers
• Length 40 Km
• Power loss per Km .25dB/Km
• Delay per Km 3.3 us/Km (velocity c)
• Power Amplifiers
• Power gain 10 dB

• 2x1 Coupler
• Coupling loss .17 dB
• 1x2 Splitter
• Coupling loss .17 dB
• Two Tunable Receivers
• Receiver One
• Wavelength 1530 nm for first
• 10 DSs then switch to 1560 nm
• Input coupling loss .17 dB
• Power threshold level 17 dBm
• Receiver Two
• Wavelength 1560 nm for first
• 10 DSs then switch to 1530 nm
• Input coupling loss .17 dB
• Power threshold level 17 dBm

29
Preliminary Modeling
ID
? (nm)
Receiver 1
Receiver 1
Receiver 2
Receiver 2
Cycles
Cycles

• Wavelength Select Test Results
• Receiver 1 receives 10 packets on 1530 nm, then
  receives on 1560 nm
• Receiver 2 receives 10 packets on 1560 nm, then
  receives on 1530 nm

30
Preliminary Modeling

• Preview of Future Modules
• 4x1 MUX, 1x4 DMUX
• OADM
• Tunable Laser
• Star Coupler
• Encoder/Decoder
• Serializer/Deserializer
• Optical Switch
• Selected L2 Protocols
• More to come

31
Potential Applications

• Flexible tool amenable to broad range of
  applications
• Networks, architectures, protocols, services,
  traffic, topologies
• Investigate tradeoffs in advanced networks
• Functionality, timing, cost
• Performance, scalability, QoS
• Fault tolerance, security
• Study transition paths from current to future
• Next-generation avionics systems
• Enabling, emerging technologies
• Bridging between networks
• Safety-critical and non-critical networks
• High-speed and low-speed networks
• Wired and wireless networks
• Passive and active networks
• Networks, interconnects, and backplanes

32
Conclusions

• Rapid virtual prototyping of high-speed optical
  networks
• Investigate tradeoffs in complex networks and
  systems
• Computer simulation
• Supported by analytical and experimental elements
• Development and refinement of key tools
• Commercial simulation tool is basis MLDesigner
• New component, subsystem, and system models
  underway
• Highly flexible and extensible environment
• Leverage other activities for model exchange and
  interoperation
• e.g. TCP, UDP, IP, 802.11, 802.3
• Broad range of applications
• Primary application is advanced avionics networks
• Strong potential in many other areas of data and
  computer communication and computation