Virtual Prototyping of WDM Avionics Networks

1
Virtual Prototyping of WDM Avionics Networks

• Casey B. Reardon, Ian A. Troxel,
• and Alan D. George
• HCS Research Laboratory
• University of Florida

2
Outline

• Introduction
• LION Library
• Modules and Components
• Modeling Environment
• Avionics System Case Study
• Traffic Characterization
• Ethernet Systems
• Candidate WDM Systems
• Simulations Results
• Message Statistics
• Analysis
• Conclusions

3
Introduction

• Optical Network Technology is Rapidly Maturing
• Increased reliability, performance,
  cost-effectiveness of optical components
• WDM an Attractive Option for Next-Generation
  Avionics Network
• Huge bandwidth capacities, better weight
  characteristics, protocol transparency, etc.
• As Optical Components Advance, Additional System
  Architectures Become Feasible
• Analysis and comparison of various architectures
  needed before an optimized architecture is
  designed
• Importance of Simulation-based Virtual
  Prototyping
• Allows timely evaluation of wide range of
  architectures
• Rapid development of various system models from
  software libraries
• Cost-effective testing of preliminary and/or
  detailed architecture designs
• Evaluation of architectures before production of
  physical prototypes
• Evaluate designs at all layers of network
  architecture
• Increased likelihood of success at system
  fabrication time
• Powerful, Flexible Tool is Needed to Aid in
  System Design Phases

4
Lion Library Overview

• LION Library for Integrated Optical Networks
• Bridge gap between optical-centric and
  network-centric modeling and simulation analysis
  tools
• Library Currently Contains 39 Optical Modules
• Couplers, splitters, lasers, receivers, etc.
• Parameters model key timing and physical
  component effects
• Low-level components used to realize any number
  of higher-level modules
• Library being updated to reflect current
  technology trends
• Integrates with variety of network modeling
  components
• Easy interface with higher-level network protocol
  models

Optical Fiber Model in LION
OADM Model in LION
5
Lion Library Overview

• Project Began with Survey of Modeling
  Environments
• Tradeoffs between optical modeling tools and
  network modeling tools
• Solution needed to combine benefits of network-
  and optics-oriented simulation tools
• MLDesigner Selected as Preferred Simulation
  Modeling Tool
• Discrete-event simulation environment, developed
  by MLDesign Inc.
• Advantages offered by MLD
• Models are fully extendible and user-definable
• Inherent hierarchical design facilitates modeling
  at multiple levels
• Supports different modeling domains with high
  fidelity
• Wireless, optical, electrical, satellite,
  time-triggered systems, unified
• Considerable network-modeling experience at UF
  with MLD and predecessor (BONeS)
• Ethernet, InfiniBand, SCI, RapidIO, Myrinet,
  Fibre Channel, LION
• Industry acceptance and technology support
• Aerospace Corp, Agere, AIRBUS (Germany), Apple,
  Astrium, Ericsson, ifEN (Germany), Honeywell,
  Infineon, KPN (Netherlands), Lockheed Martin,
  Motorola, Philips, Rockwell Collins, Siemens,
  etc.
• gt40 Universities
• Interoperability with other tools
• SatLab
• MATLAB/Simulink

6
Avionics System Case Study

• Representative Network Nodesand Traffic to Test
  Models andDesigns
• Sample Network Parameters of Contemporary
  Avionics
• Traffic data provided by Rockwell Collins
• 28 nodes connected to network,in 7 different
  subsystems
• Traffic classified into 39 VLIDs, with 3 priority
  levels
• Traffic types vary from periodic,to random, to
  bursty
• Source data rates range from 10 Kb/s to 250 Mb/s
• Aggregate throughput of systemaverages 1.27 Gb/s

Aircraft LAN Source Traffic Patterns
7
Avionics System Case Study

• Baseline System Constructed with Switched
  Ethernet
• 3 16-Port IP/Ethernet switches
• Switches contain cast andgigabit speed links
• Each VLID associated w/ amulticast address and
  QOSlevel
• Switches QoS-enabled to handle prioritized
  traffic

• Two Candidate Optical WDM Systems Considered
• Both WDM systems employ U-shaped bus as backbone
• Single fiber media used to share all network
  traffic
• Network is entirely passive

8
Avionics System Case Study

• WDM Nodes Limited to Fixed-Wavelength Lasers and
  Receivers
• Lasers operate at 1 Gb/s transmission rates
• Traffic Divided Between 7 Wavelengths for Each
  System
• TDM on each wavelength to govern multiple access
  between nodes
• Larger proportion of TDM slots assigned to
  high-priority traffic
• Additional wavelengths reserved for control and
  monitoring signals
• WDM System A
• Each node outfitted with single fixed-wavelength
  laser
• One wavelength reserved for each of 7 subsystems
  of nodes
• Most traffic sent and received within a subsystem
• Up to 5 fixed-wavelength receivers needed in
  nodes
• Additional receivers needed for any traffic
  received from separate subsystems
• WDM System B
• Each node has 2 fixed-wavelength lasers
• Up to three optical filters needed at each node
• Wavelength assignment based on VLID groups
• Most inter-subsystem traffic shared on same
  wavelength

9
Simulation Results

• Results Collected From Ten Seconds of Simulated
  Traffic
• Both WDM Systems Far Outperform the Baseline
  Ethernet
• Average message latencies are far higher, for all
  traffic classifications
• In-transit packet processing not required in
  optical systems
• All Systems Capable of Fast Transfer of
  High-Priority Traffic
• Lower-priority traffic suffers due to QoS demands
  in Ethernet
• TDM/WDM Offers Deterministic Network Performance
• Overall latency deviation is orders of magnitude
  smaller with WDM
• Ethernet suffers from contention at heavily
  congested ports, unlike optical systems
• Adding additional switches would do little to
  remedy this effect

Table 1 Summary of Case-Study Results
10
Simulation Results

• Comparison of WDM Candidate Architectures
• System B delivers 85 of packets under 150 us,
  20 more than System A
• Leads to lower average latency and standard
  deviation of latencies in system B

• Additional Laser Transmitter Offers Several
  Benefits to Candidate WDM Avionics Architecture
• Better traffic load-balancing capable within
  wavelengths
• Especially important for high-priority traffic,
  which used extra time-slots to ensure faster
  transfers
• Transmitter hardware requirement may be offset by
  fewer receivers
• Important cost-benefit option to be considered by
  system designers

11
Further Results Discussion

• Large Potential Within Candidate WDM
  Architectures
• Systems are highly scalable
• Faster transmitters available than 1 Gb/s used in
  these models
• Additional wavelengths provide additional
  bandwidth
• 8 wavelengths is very modest for modern WDM
  systems
• Room for smarter access protocols to enhance
  performance
• Pre-assigned, static time slots cannot adapt to
  different traffic demands
• Reservation schemes could handle changing traffic
  patterns
• Case-Study Is Just One Example of LIONs
  Capabilities
• WDM candidate architectures just a beginning
  sample
• The U-bus can easily be modified to create a ring
  or other simple topologies
• Optical routers and switches allow WDM-based
  meshes and other more advanced topologies
• Library captures measurements beyond just
  message latencies
• System-wide estimates of weight, cost, power
  requirements
• Optical power-budget monitoring and analysis

Topology Illustrations
12
Conclusions

• Demonstration of LION To Model and Evaluate
  Optical WDM LAN Networks for Avionics
• Rapid virtual prototyping provides cost-effective
  method of system evaluation
• Comparison of Ethernet and 2 WDM Architectures
• WDM offers significant advantages in terms of
  performance, scalability, weight, and protocol
  independence
• Multiple wavelength transmitters can improve
  performance by enabling better traffic
  load-balancing
• Wide variety of architectures and design options
  can be analyzed using LION
• Easily adapted to evaluate systems carrying
  multiple protocols, and/or mixed digital and
  analog signals
• LION Not Limited to Analysis of Networks For
  Avionics
• Future Work
• Phase-1 STTR funded by Naval Air Systems Command
• Identify and evaluate promising WDM LAN
  candidates
• Enhance LION to include modern and future WDM
  components
• Rugged tunable lasers, switches, etc.
• Increased fidelity of impact from physical and
  environmental effects
• Parameters to be acquired through high-fidelity
  physical component simulation outside of LION
• Incorporate aspects of fault-tolerance and
  alternate control and legacy protocols with
  optical components
• Investigate candidates and options to support SAE
  standard under development