Virtual Prototyping ofHigh-Performance Optical Networks for Advanced Avionics Systems: Virtual Prototyping of High-Performance Optical Networks for Advanced Avionics Systems Dr. Alan D. George Ian Troxel, Ramesh Balasubramanian Chris Catoe, Jeremy Wills High-performance Computing and Simulation (HCS) Research Laboratory Department of Electrical and Computer Engineering University of Florida


Outline: Outline Introduction Tool Evaluations MLDesigner Overview Optical Component Modeling Advanced Avionic System Models Switched Aircraft LAN Switchless Pixel Bus Switchless Aircraft LAN Conclusions


Introduction: Lab Research Areas high-performance computer networks high-performance computer architectures parallel and distributed computing reconfigurable and fault-tolerant computing Lab Research Methods modeling and simulation testbed experimentation software design andamp; development hardware design andamp; development Introduction


Introduction: Introduction Team Modeling Experience Network modeling SCI and SCI/RT net. (BONeS, UltraSAN) Myrinet network (BONeS) Fibre Channel network (BONeS) Architecture and systems modeling RISC (BONeS, MLD) CMP (C, extended SimpleScalar) SMP (BONeS) Reconfigurable network proc. (BONeS) HWIL and SWIL simulation (BONeS) New efforts underway Optical avionics networks (MLDesigner) – project focus Fast andamp; accurate simulator for adv. HPC clusters and grids (MLD) Performance andamp; dependability sim. for mission assurance (MLD) FPGA-based reconfigurable architectures and systems (MLD) Educational tool for grad. courses in networks andamp; systems (MLD)


Slide5: Cluster-based lab HPC grid Collection of 11 PC clusters 480 Pentium-compatible CPUs Newest: Xeon (40) and Opteron (32) 308 networked Linux nodes 102 GB memory, 5.18 TB storage PCI64/66 and PCI-X support Other compute resources ES80 AlphaServer ('Marvel') Cluster of Sun workstations Networking testbeds 5.3 Gb/s Scalable Coherent Int. 10 Gb/s InfiniBand (4X) 1 Gb/s Gigabit Ethernet (Fiber/UTP) 10 Gb/s Ethernet (Beta test) 1.28 Gb/s Myrinet, 3.2 Gb/s QsNet 1.25 Gb/s Giganet cLAN Introduction Facilities for Simulative and Experimental Research


Tool Evaluations: Tool Evaluations Initial Goals for Optical Networking Tool Model networking issues (data-link layer, network layer, etc.) while also achieving realistic representation of optical physical layer Appropriate level of speed vs. fidelity Library of pre-built models Stability and maturity Responsive technical support Ease of use Interoperability Cost-effective


Tool Evaluations: Tool Evaluations Divergent Roads Networking tools Protocols and topology focus Typically open source Physical layer abstracted Optical-layer tools Optics focus Typically expensive No networking protocols Others Blank-canvas approach Various strengths and weaknesses


Tool Evaluations: Tool Evaluations MLDesigner selected as best all-around tool Flexible Models fully extendible and user-definable Supports different modeling domains with high fidelity Wireless, optical, electrical, satellite, time-triggered systems unified In 2003, HCS lab built Lib. for Integrated Optical Networking (LION) Supports software/hardware in-the-loop simulation (Berkeley Sockets) Industry acceptance and technology support Aerospace Corp, Agere, AIRBUS (Germany), Apple, Astrium, Ericsson, ifEN (Germany), Infineon, KPN (Netherlands), Lockheed Martin, Motorola, Philips, Rockwell Collins, Siemens, large US semiconductor manufacturer, large US aerospace company, etc. andgt;40 Universities Cost effective $7-9K annual corporate license (per seat); free for universities Interoperability with other tools SatLab MATLAB/Simulink


Tool Evaluations: Tool Evaluations Advantages of MLD+LION over Networking Tools (ex. OPNET Modeler) *Key advantage


Tool Evaluations: Tool Evaluations Advantages of MLD+LION over Optical-Layer Tools (ex. VPI Systems) *Key advantage


Component Modeling: Component Modeling Optical components easily replicated with different parameter settings to produce commercial product models (ex. Genoa GT111 Amplifier)


Switched Aircraft LAN: Switched Aircraft LAN Note: Demo after presentation Key tradeoffs QoS thresholds and algorithms Latency / Bandwidth Power analysis Cost analysis (baseline cost) Key features Baseline system Application and traffic study Virtual Links with QoS TCP / IP / Ethernet systems Fiber and copper links


Switchless Pixel Bus: Switchless Pixel Bus Note: Demo after presentation Key tradeoffs Latency / Bandwidth Optical power budget Electrical power analysis Cost analysis (baseline cost) Key features 4 channels @ 2.5Gbps (10Gbps aggregate) Fixed optical components (cheaper) TDM System


Switchless Pixel Bus: Switchless Pixel Bus Key features 4 channels @ 10Gbps (40Gbps aggregate) 4 independent wavelengths (better security) More optical components (increased cost) TDM system provides cost-effective solution if bandwidth limitation is sufficient WDM system provides better bandwidth and security at increased cost in both $ and power Key tradeoffs Latency / Bandwidth Optical power budget Electrical power analysis Cost analysis (baseline cost) Note: Demo after presentation WDM System


Switchless Aircraft LAN: Switchless Aircraft LAN Key tradeoffs WDM / TDM Compare to baseline architecture Latency / Bandwidth Power analysis (mostly passive) Cost analysis Design in progress Key features Candidate system Unified bus Switchless Tunable wavelengths Optical switching Increased reliability Supports bandwidth growth Note: Similar trade study to be performed upon model’s completion Gateway Legend G# = Gigabit Ethernet FE = Fast Ethernet TT = Time Triggered


Potential Applications: Potential Applications Flexible tool amenable to broad range of applications Networks, systems, architectures, protocols, services, traffic, topologies Investigate tradeoffs in advanced networks Functionality, timing, cost Performance, scalability, QoS Fault tolerance, security Investigate transition paths 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


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


Backup Slide : Backup Slide 1985 1990 1995 2000 F U N C T I O N A R C H I T E C T U R E 1st Generation 2nd Generation 3rd Gen. System Design Market Evolution


Backup Slide: Backup Slide


Backup Slide: Backup Slide MLD Architecture Selected Customer Applications: MIL wireless communications network MIL satellite systems design, wireless communication design Development of next-generation GPS satellite system Integrated radar system Design of ADS-B based aircraft traffic management system New rapid design flow for computers Mission-level design of unmanned underwater vehicle (UUV) Mission, sensor, Gandamp;C, communication system design Each switched from BONeS, COSSAP, OPNET, Matlab/MatrixX, SPW, C/C++