FiberO ptic R eceiver A nalog to D igital C onverter D igitalSignal Processing FiberO ptic Splitter FiberO ptic C oupler FiberO ptic C oupler FiberO ptic Splitter David Sheets 343 Layth Al-Jalil 335 Adam Fritz 309 Jay Becker303 FiberOpticCable $50 FiberOpticSplitters, Couplers,etc. $500 Poster $50 NIC(s)and Tranceivers $5,000 Student Labor $12,790 Fiber optic network in ring Fiber optic network in ring topology topology Custom software implementing a Custom software implementing a Time Division Multiplexing (TDM) Time Division Multiplexing (TDM) scheme scheme Documentation summarizing Documentation summarizing conclusions conclusions Engineers at Lockheed Martin & Engineers at Lockheed Martin & researchers at Iowa State researchers at Iowa State University University Provide a low latency medium for Provide a low latency medium for transmission of high resolution transmission of high resolution video and other high bandwidth video and other high bandwidth data data Test the viability of different Test the viability of different protocols, processor protocols, processor configurations, and future fiber configurations, and future fiber optic technology solutions optic technology solutions Scheduler N ode 1 N ode 3 N ode n B ack up N ode 2 May 07-06 Team Members: Layth Al-Jalil (Cpr E) Adam Fritz (EE) Jay Becker (Cpr E/ComS) David Sheets Faculty Advisors: Dr. Mani Mina Dr. Arun Somani Dr. Robert Weber Introduction Project Schedule Problem Statement & Solution Operating Environment There is no requirement to There is no requirement to develop a custom network develop a custom network interface card interface card Development of this system will Development of this system will be largely drawn from existing be largely drawn from existing research research Budget of $5000 Budget of $5000 Project lifespan is 9 months Project lifespan is 9 months COTS equipment must be primarily COTS equipment must be primarily used used Assumptions & Limitations End Product Project Requirements Topology will support bi-directional signal propagation Topology will support bi-directional signal propagation Scheduling software will run on NIC independent of host Scheduling software will run on NIC independent of host Network will tolerate insertion and removal of nodes Network will tolerate insertion and removal of nodes Design must be a cost effective solution Design must be a cost effective solution $5000 budget does not support an extensive system $5000 budget does not support an extensive system Range of commercially available fiber optic transceivers and Range of commercially available fiber optic transceivers and attenuation through topology limits the network to four nodes attenuation through topology limits the network to four nodes Software must run on the NIC, not host platform Software must run on the NIC, not host platform Power loss across node sub-networks limits total number of nodes Power loss across node sub-networks limits total number of nodes Estimated Cost for Spring 2007 Estimated Cost for Spring 2007 ($18,390) ($18,390) Research will be conducted to determine similar industry Research will be conducted to determine similar industry solutions solutions Consultation with customer to clarify requirements Consultation with customer to clarify requirements Consultation with advisors and graduate students on technical Consultation with advisors and graduate students on technical issues issues Subsystem prototyping and testing used to guide final Subsystem prototyping and testing used to guide final integration integration The performance of high speed fiber optic systems is interlaced with The performance of high speed fiber optic systems is interlaced with the issues of network topology, fault tolerance, and decentralized the issues of network topology, fault tolerance, and decentralized control. Our team is building a fiber optic network supporting ten control. Our team is building a fiber optic network supporting ten gigabit per second baud rate that utilizes a ring topology and bi- gigabit per second baud rate that utilizes a ring topology and bi- directional data transfer to provide a fault tolerant network directional data transfer to provide a fault tolerant network solution. This design enables every node in a network to serve as a solution. This design enables every node in a network to serve as a scheduler, thus providing decentralized control where the loss of a scheduler, thus providing decentralized control where the loss of a control node is mitigated by having immediate back up available. When control node is mitigated by having immediate back up available. When complete, this solution will be directly integrated into conventional complete, this solution will be directly integrated into conventional avionics architectures. avionics architectures. Abstract Avionics platforms are increasingly demanding greater throughput Avionics platforms are increasingly demanding greater throughput between system elements. These requirements are driven by the need to between system elements. These requirements are driven by the need to transmit real time video to the pilot and crew, broadcast complex transmit real time video to the pilot and crew, broadcast complex tactical data throughout a military vehicle, and to provide expansion tactical data throughout a military vehicle, and to provide expansion bandwidth for the next generation of equipment . This team is realizing bandwidth for the next generation of equipment . This team is realizing that vision as a fiber optic network because as data transfer moves that vision as a fiber optic network because as data transfer moves beyond the ten gigabit per second rate the only effective, EMI beyond the ten gigabit per second rate the only effective, EMI insensitive medium is fiber optics. insensitive medium is fiber optics. Estimated Resources Approach and Considerations Closing Summary Modern avionic platforms require data networks that can accommodate Modern avionic platforms require data networks that can accommodate current and future bandwidth needs current and future bandwidth needs Industry trend is for many modules to share in tasks of processing Industry trend is for many modules to share in tasks of processing data data Build a network capable of 10 Gbps transceiver to transceiver Build a network capable of 10 Gbps transceiver to transceiver communication communication Client: Military avionic platforms and Military avionic platforms and derivative systems derivative systems High Speed Communication and High Speed Communication and Dependable Computing Dependable Computing Laboratories Laboratories Intended Users and Uses Estimated Personnel Hours Estimated Personnel Hours (1,290 Total Hours) (1,290 Total Hours) Department equipment was utilized for initial testing Department equipment was utilized for initial testing None of the available equipment supports 10Gbps None of the available equipment supports 10Gbps Multimode equipment is available in the lab; single mode equipment is not Multimode equipment is available in the lab; single mode equipment is not Concepts must be proven with lab equipment before making purchases Concepts must be proven with lab equipment before making purchases TDM/Loop Topology Bi- directional data flow N ode 1 C oupler/ Splitter C oupler/ Splitter C oupler/ Splitter C oupler/ Splitter Tx Rx N ode 1 N ode 1 N ode 1 Tx Tx Tx Rx Rx Rx Bi-directional Data Flow Input / Output NIC Transmitter NIC Receiver TDM Order of Transmission Node n to transmit (Represents order of control, not signal path) Node 2 to Transmit Node 3 to transmit Design Objectives Functional Requirements Design Constraints Measurable Milestones Proposed Approach Technologies Considered Testing Considerations Fiber optic transceivers will transmit and receive serial data at Fiber optic transceivers will transmit and receive serial data at a 10 Gbps a 10 Gbps Each node will distinguish between a strong signal and a weak Each node will distinguish between a strong signal and a weak signal signal Scheduler will determine the order and duration of each node’s Scheduler will determine the order and duration of each node’s transmission transmission Each node will transmit when the scheduler transmits authorization Each node will transmit when the scheduler transmits authorization Measure power output from a dummy fiber optic network to confirm Measure power output from a dummy fiber optic network to confirm that signals propagate in the design that signals propagate in the design Implement a two node network to test the TDM scheduling algorithm Implement a two node network to test the TDM scheduling algorithm Expand network to four nodes and test TDM scheduling algorithm Expand network to four nodes and test TDM scheduling algorithm Upgrade network from 2.5 Gbps to 10 Gbps and measure bit rate Upgrade network from 2.5 Gbps to 10 Gbps and measure bit rate Single mode technologies chosen for ease of use (fiber, couplers, Single mode technologies chosen for ease of use (fiber, couplers, splitters, etc.) splitters, etc.) Myrinet supports 10Gbps but software is not customizable, thus it was Myrinet supports 10Gbps but software is not customizable, thus it was rejected rejected Xilinx Virtex IV supports 10Gbps and is reprogrammable Xilinx Virtex IV supports 10Gbps and is reprogrammable Optical signal up to 10Gbps Electrical RF signal Digital signal Processed digital signal