Special Workshop: DARPA SC2

Wednesday, 15 November






SC2 Program Overview


  • What is Collaborative Intelligent Radio Network (CIRN) technology
  • Why is SC2 important
  • Key tech and program challenges
  • What is Colosseum
  • Program structure and progress to date

Paul Tilghman, DARPA

Collaborative Intelligent Radio Networks as a Multiagent System


  • CIRN as a multi-agent AI problem
  • Competition as driver for multi-agent AI technology
  • Strategic agents and incentives for collaboration

Michael Wellman, University of Michigan

Morning Session

WinnForum Chair Introduction


How SC2 topics tie into the WinnForum Top 10 list

Manuel Uhm

Technology Challenges


1. Application challenges

  • What are the characteristics of a use case or scenario in which collaboration offers benefits?
  • What canonical cases are good for driving research?

John Chapin, DARPA SETA

2. New wireless network design challenges

  • What new capabilities are needed in PHY/MAC/NET?
  • What new information has to be created, disseminated, or processed?

Jody Neel, Federated Wireless

The Colosseum System


3. MCHEM: At-scale realistic testing in a channel emulator

  • Hardware and software architecture
  • Why is 256x256 not just a linear scale-up of COTS 16x16 emulators?

Ashish Chaudhari, Ettus

4. Colosseum: A remotely accessible, shared test resource

  • Hardware and software architecture surrounding the MCHEM
  • What is hard about efficient sharing of a large channel emulator among remote users.
  • Operational experience: challenges and fixes

Jason Uher, Johns Hopkins APL

5. Scenarios for testing Colosseum and CIRNs

  • Challenges and solutions in generating realistic tests
  • The scenario compilation pipeline
  • Test and validation methods for Colosseum itself

Binoy Kurien, MIT Lincoln Labs

Afternoon Session 1

Collaboration Protocol


6. SC2CP: The key enabler for collaboration

  • What are the concepts that CIRNs need to express?
  • What are collective actions or decisions we would like CIRNs to do? Which have occurred in Phase 1 so far?
  • How can a program like SC2 evolve a useful protocol?

Craig Pomeroy, DARPA SETA

7. A collaborative AI perspective on SC2CP

  • How does SC2CP compare to protocols used in other collaborative AI domains?
  • What are the fundamental problems in protocols like this?
  • What approaches have proven broadly useful that should be applied in this domain?

Andrew Adams, Johns Hopkins APL

Early Experiences


8. Observations from SC2 Phase 1

  • What has been learned from scrimmages held to date?
  • What important challenges are competitors struggling with?

Paul Tilghman, DARPA

Transitioning Collaborative Spectrum Sharing from Research to Reality


Introduction: The SC2 Stakeholders Council

Paul Tilghman, DARPA

Panel Discussion

  • What mission needs of my organization could be helped by collaborative spectrum sharing?
  • What has to be done to the technology post-competition to make it fit those needs?
  • What needs to be proven to get me/my organization seriously interested in adoption?
  • What might deployment look like from a Policy perspective (e.g. how to certify a network that learns)?

Manuel Uhm (Moderator)

Ravi Potluri, Verizon

Ali Khayrallah, Ericsson

Kenneth Baker, NTIA ITS

Next Steps


  • Program and CIRN technology roadmap
  • How interested parties can get involved

Paul Tilghman, DARPA

Afternoon Session 2



  • Colosseum: schedule, configure, run an experiment
  • CIRNs: run a scrimmage match
  • Spectrum visualization and Collaboration Protocol logs: investigating what happened

Jason Uher, Johns Hopkins APL

Q&A Discussion


Manuel Uhm, Moderator

powered by MemberClicks