User and Network Interplay in Internet Telemicroscopy Prasad

User and Network Interplay in Internet Telemicroscopy Prasad Calyam (Presenter) Nathan Howes, Mark Haffner, Abdul Kalash Ohio Supercomputer Center, The Ohio State University IMMERSCOM, October 11 th 2007

Topics of Discussion § Telemicroscopy Overview § Motivation § Use-cases § Solutions § Telemicroscopy Session Model § User and Network Interplay § Testbed for Experiments to Characterize Model Parameters § Performance Analysis § OSC’s Remote Instrumentation Collaboration Environment (RICE) § Features § Demo Video § Conclusion 2

Telemicroscopy Overview § Academia and Industry use computer-controlled scientific instruments § Electron Microscopes, NMR, Raman Spectrometers, Nuclear Accelerator § For research and training purposes § Cancer Cure, Material Science, Nanotechnology § Instruments are expensive ($450 K - $ 4 Million) and need dedicated staff to maintain +) Remote instrumentation benefits § Access to users who cannot afford to buy instruments § Return on Investment (ROI) for instrument labs § Avoids duplication of instrument investments for funding agencies (NSF, OBOR) § Useful when physical presence of humans around sample is undesirable Telemicroscopy is remote instrumentation of electron -) Remote instrumentation drawbacks microscopes 3

Telemicroscopy Use-cases § Tele-observation versus Tele-operation 4

Telemicroscopy Solutions § Hardware-based: KVM over IP (KVMo. IP) § Encoder-Decoder pair for frame-differencing based video image transfers § Pros: High quality video and optimal response times § Cons: Expensive, Special hardware and high-end bandwidth requirements § Software-based: VNC – remote desktop software § Raw or copy-rectangle or JPEG/MPEG encoded video image transfers § Pros: Inexpensive, Easily deployable § Cons: Improper PC hardware or network congestion can degrade video quality and optimal control response times 5

Related Work § Telemicroscopy over Internet 2 § Gemini Observatory § Nano. Manipulator § Telescience Project – National Center for Microscopy and Imaging Research, UC San Diego § Ultrahigh Voltage Electron Microscope Research Center – Osaka University § Common Instrument Middleware Architecture (CIMA) – Indiana University +) Novel applications for controlling instruments +) All said “it works” over XYZ network paths and listed challenges they overcame -) None have quantified § Tele-presence Microscopy – Argonne National Lab’s Advanced Analytical performance in Electron Microscope facility terms of network Study Motivation: Understanding User and Network interplay can effects help us improve reliability and efficiency of Telemicroscopy and thus 6 -) None have deliver optimum user Qo. E

Telemicroscopy Session Model (a) Session Model Parameters during a functions (b) Closed-loop Control System Representation → user-activity (key strokes and mouse clicks) session involving n microscope → average video image transfer rate at the microscope end → network connection quality → input-output scaling factor; unique to a microscope function → seed image transfer rate; for quick screen refresh → average video image transfer rate at the user 7

Telemicroscopy Session Model (a) Session Model Parameters (b) Closed-loop Control System Representation (c) Transfer Function (d) End-user Qo. E relation in a Telemicroscopy session § Demand – Effort the user had to expend to perform n actions § Supply – Perceivable video image quality during the 8

Telemicroscopy System States (Effects of H parameter) (a) State Transitions (b) System Supply-Demand Performance 9

Case Study: OSC Collaboration with OSU CAMM § OSU Center for Accelerated Maturation of Materials (CAMM) has acquired high-end Electron Microscopes § Used for materials modeling studies at sub-angstrom level § OSC providing systems and networking support for Telemicroscopy § OSCnet supporting end-to-end bandwidth requirements § Image processing of samples (automation with MATLAB) for Analytics service § Telemicroscopy Demonstrations § Supercomputing, Tampa, FL (Nov 2006) § Internet 2 Fall Member Meeting, Chicago, IL (Dec 2006) § Stark State University/Timken, Canton, OH (Mar 2007) 10

Telemicroscopy Testbed § Experiments to characterize session model parameters § Test cases with different network connections – CAMM requirements § (a) 1 Gbps LAN (Direct connection to Users in neighboring room) § (b) Isolated LAN (Users in the same building ) § (c) Public LAN (Users in different buildings on campus) § (d) WAN (Users on the Internet) § Performance analysis goals § Bandwidth, latency and packet loss levels for optimum user Qo. E § Traffic characterization for studying inter-play between 11

WAN Testbed (a) Setup (b) WAN Path Performance 12

Performance Measurements Collected § End-user Qo. E Measurements (Subjective Metrics) § § § Mean Opinion Scores (MOS) of “Novice” and “Expert” Users Time for completion of “basic” and “advanced” Telemicroscopy tasks by Novice and Expert Users Network Measurements (Objective Metrics) § Collected using Ethereal/TCPdump and OSC Active. Mon 13 § Metrics: Data rate, Protocols Summary

Network Connection Quality (ψnet) and User Qo. E (qmos) § qmos notably decreases with decrease in network connection quality § User Qo. E is highly sensitive to network health fluctuations § Novice more liberal than Expert § Time taken to complete a task increases with decrease in network connection quality NOTE: qmos of 5 corresponds to “at the microscope” Qo. E 14

Network Connection Quality (ψnet) and User Control (bin) § Mouse and Keyboard traffic is TCP traffic § Higher TCP throughput on poor network connections Increased user effort with keyboard and mouse on poor connections § “Congestion begets more congestion” 1 Gbps LAN – Expert Task-1 60 B/s Task-2 Task-3 Public 100 Mbps LAN – Expert Task-1 900 B/s Task-2 Task-3 100 Mbps WAN – Expert Task-1 Task-2 Task-3 1400 B/s 60 s User expends minimum effort with keyboard and mouse to complete use-case 100 s User expends notably more effort with keyboard and mouse to complete use-case 15 140 s User expends a “lot” of effort with keyboard and mouse to complete use-case

Network Connection Quality (ψnet) and Image Transfer Rate (Δbout) § “At the microscope” Qo. E requires ~30 Mbps between user and microscope ends § Other WAN tests at SC 06 (Tampa) and Internet 2 Fall. MM (Chicago) to microscopes at CAMM (Columbus) § Usable on ~(10 -25) Mbps WAN connections § Usable if one-way network delays within ~50 ms; as much as ~20% UDP packet loss tolerable if adequate bandwidth provisioned 16

OSC’s Remote Instrumentation Collabration Environment (RICE) § Leverages our user and network interplay studies for “reliable” and “efficient” Telemicroscopy sessions and thus delivers optimum user Qo. E § Customizable software on custom server-side hardware for Telemicroscopy § Best of VNC and KVMo. IP worlds § RICE Features § Network-aware video encoding § Optimizes frame rates based on available network bandwidth § Manual video-quality adjustment slider § Network-status and user-action blocking § Warns user of network congestion that leads to unstable session state § Blocks user-actions during extreme congestion scenarios and prevents breakdown § Collaboration tools § Vo. IP, Chat, Annotation, Command-abstraction § Multi-user support § Control-lock passing, collaborators presence, colored-text chat conference § Workflow and Image management 17

RICE Demo Video 18

RICE use-cases for online learning § Remote students can view instructor (also remote!) controlling different types of scientific instruments § Efficiently – with the appropriate video frames to match last mile network capabilities § Reliably – without worrying about damaging the instrument § Multi-party Vo. IP and Chat collaboration § Image Annotation § Instructor can pass control to students - train them to operate the instrument during the class § Students can conduct lab sessions at their assigned slots on the instruments § Students image files can be organized and hosted at a central server § Analytics can be supported using a web-service to analyze the image data sets 19

Future Work § Shared instrumentation uses OSC’s state-wide resources § Networking, Storage, HPC, Analytics Cyberinfrastructure for Shared Instrumentation 20

Shared Instrumentation @ OSC § Plans underway to support shared instrumentation for § Ohio State University: CAMM Electron Microscopes, Chemistry Department Spectrometers and Diffractometers, Astronomy Department Telescopes § Miami University: Electron Microscopes, EPR Spectrometers § Ohio University: Nuclear Accelerator 21

Thank you for your attention! ☺ Any Questions? 22