Keynote: Dave Goodman (Program Director at NSF)

21st Century Networking: Beyond Bits and Packets

• NSF GENI Initiative
  • Promote large scale internets
  • reinvent the internet
• Facility for rsearch
  • open
  • large scale
  • realistic
• Basic questions
  • What ar rquiremnts for globa network of 21st century?
  • How would we design tomorrow's global ntwork from scratch?
  • Needs to be address by experiments at a large scale
• Why a new network?
• A few examples:
  • Security
  • Resilience
  • ~Mobility~
  • New technology
  • New paradigms
  • Management
  • Ease of use
• Conception-to-deplyment

       models   code      r------>Deployment

Analysis->Sim/Em->Experiment (feedback via measurements) -> Analysis ^--New Ideas <---^Results(back to analysis)

• Fill in the gap between small scale testbeds and shared deployable

infrastructure

• How will GENI work
  • Slicing
  • Virtualization
  • Programmability
• Overview
  • National Fiber Infrastructure
  • Edge Clusters
  • POPS
  • Programmable Routers
• GENI Mgmnt Core
  • namespace for users, slices, components
  • set of interfaces
  • support for federation
  • security
  • http://geni.net
• Geography of information
  • Where? (Local, Regional, Global)
  • When is it relevant? (Briefly? For a while? Long term?)
  • When is it needed? (Conversational? Interactive? Streaming? Background?)
  • Who wants it? (One person? Group? Millions?)
  • How is it conveyed? (Sounds? Pictures? Text? Video?)
• Use geography and intended scale to tailor delivery approach
• Theory is important, but make sure you link it to the physical world

New Yorker Cover Photo: "A Higher Calling" (Aug 3, 1998)

• Audience questions
  • What about breakthrough discoveries? Will these be able to be incorporated?

How much flexible is there really?

• You don't have to know anything to build a testbed

Session 1: Testbeds

A Software Architecture for Physical Layer Wireless Network Emulation

• Wireless networks have fundamental difference: physical layer coupled to all

layers

• Realism vs. control
• Testbed limitations
  • Distributed uncontrolled signal propagation environment
  • Controlled but highly-constrained signal propagation environment
  • Often cumbersome (vehicles)
• Key idea
  • Virtualized signal propagation environment
  • Everything is real except for physical layer
• Emulation overview
  • Signal copying and attenuation and re-summing
  • Radio reception is real
• Getting Signal Propagation Info
  • Modeling
  • Recording
  • Ray Tracing
• Planning to allow remote access

An Integrated Approach to the Development of Wireless Network Protocols

• Questions
  • Is channel condition done in software?
    • Yes, in real time

Adaptive Wireless Network Testbed for Cognitive Radio Technology

• Goals
  • testbed that models actual PHY layer behavior, and is scalable
• Problems with CR testbed
  • Allocation of real devices in not flexible or scalable
  • PHY model is usually too simplified, so many phenomena like prop. or

interference are not considered.

• Emulation vs. Simulation
  • Em: Lim node, real, signal strength, BER, int
  • SIm: many node, simplified model, memory consumption, queueing delay, protocol
• Main idea: Emulation and Simulation executed cooperatively
• Testbed framework
  • MIRAI-SF simulator with cog radio plugins for real devices
  • MIRAI-CREF (cog radio exec framework)
• MIRAI-SF is an open source simulator available from NiCT
• Related work
  • CRSETE - Cog Radio Environment in GENI
• Question about separation between MAC and PHY