1. GENI ‘Global Environment for Network Innovations’
Notes:
Pleased to have this opportunity to share our thinking about GENI with you.
This reflects thinking of the research community as well as thinking of CISE.
Of course, if this doesn’t come across as compelling it is my fault -- I have to work extra hard to make this case.

2. Future Internet - A Lead-up to the NSF GENI Project
Notes:
Pleased to have this opportunity to share our thinking about GENI with you.
This reflects thinking of the research community as well as thinking of CISE.
Of course, if this doesn’t come across as compelling it is my fault -- I have to work extra hard to make this case.

3. About GENI
….a continental-scale, programmable, heterogeneous, networked system driving “clean-slate” future internet / communications research
Mobile Wireless Network
Sensor Network
Edge Site
Federated International Facility

4. GENI Organizations Major Research Equipment Facility Construction (MREFC)
NSF
Program Director
GENI
Chair
Project Director
GENI Science Council
(GSC)
GENI Project Office
(GPO)
Voice of research and education community
Scientific leadership - evolution of Science Plan
Define user allocations process & criteria
Development of education & outreach plan
Research infrastructure experience – software-intensive
projects
Experience with computing community
Project management – MREFC process
GPO is at BBN
GENI Project Director is Chip Elliot

5. GENI To Date
The community has engaged in concept development since
NSF CISE funding has supported
Early concept development – GENI Planning Group
Early prototype development
Solicitation for proposals to establish GENI Project Office
GENI Science Council was established in Spring of 2007
GENI Project Office was awarded to BBN in May 2007

6. Project Life Cycle
Formal NSF process to be followed to qualify for budget consideration
Preconstruction Planning - Three formal design stages, with well-defined interfaces
Construction
Operations

7. Project Lifecycle Construction Operations CDR FDR PDR GPO Award Period
Conceptual
Design
Preliminary
Final
time
Identify Project Office
NSF will work collaboratively with the GPO and GSC to determine the dates of these reviews, required by the formal process, as the GENI community proceeeds in planning process.

8. Internet Futures – An Ecosystem
Digital Living
Edges
Networked Embedded
Critical Infrastructure
Data Grid
E-science
Language
Content Centric
Networked
Sensors
Responsive
environment
Service Oriented
Evolvability
Capabilities
Security Robustness
Mobility Ubiquity
Autonomicity
DTN
Privacy-enabling
MAN
Location-aware
Circuit
Networks
FTTH
Routed
Wireless
Switch
VPN
Packet
Wireless
Router
WDM Optical
Actuators
Sensors
Communications
Research and development for the future of the Internet takes place in an ecosystem of needs, capabilities and technology developments.
So what do we expect from the underlying network and distributed systems infrastructure?
Computing
Displays
Cluster Computing
Multicore
Radios
Optical
Storage
Memory
Devices
Antenna
PIC
Laser
Optical Fiber
Transistor
MEMS
Codes
Basic
Science
Organic
Semi
C-nanotube
Quantum crypto
Mathematics
Materials
Nanotechnology
Physics

9. GENI: The Facility Native Access for ~ 200 Universities
GENI is in Early Planning, But Some Requirements Have Become Clear
A Continent-Scale, Evolvable, Optical Substrate
Native Access for ~ 200 Universities
Native Access Will Be Considered for Non-Academic Sites
Wireless networks
Characteristics of Interest Include: Location Awareness, High Mobility, End-Users.
Support of Sensor Internet and Other Edge Research
Partnerships and Federations
Federation of GENI on International Scale Expected
We Expect This Will Associate With NSF’s International Connections Program
We Have Had Discussions About Facilities With GENI-like Qualities in the EU, Japan, Korea, China, Latin America and the United Arab Emirates
We Welcome Discussions
GENI must meet requirements of research agendas emerging from the ecosystem shown before. This research agenda is developed by the GENI Science Council. The manner in which the requirements are expressed by a facility design are determined collaboratively by the GSC and the GPO (in consultation with NSF).

10. Activities GENI Science Council GENI Project Office
Co-Chaired by Scott Shenker, UCB, and Ellen Zegura, Georgia Tech
Information in GSC area of GENI.net website
GENI Project Office
Facility Working Groups Have Open Participation – See Charters and Other Information Now Available in Working Groups Area of GENI.net website
First GENI Engineering Conference (GEC) Will Be Held October 9-11, Minneapolis – Call For Participation and Open Registration on
During the present period, NSF is pleased to have broad community participation in understanding of both the research agenda for GENI and the designs for the facility that will support this research. Working Groups at the GPO will participate in spiral design, so that their work tracks (in a disciplined way) to the requirements from the research agenda, as they become known.

11. Modalities
RFID Left Hand
Retinal Implant
WIFAN SENSOR

12. A view of innovation Design and trial without roadblocks – clean-slate
Novel distributed systems, services or support
Cross-cutting protocols
Large View
Applications
And user requirements
Current network
roadblocks
Disruptive technologies
Network capabilities
Clearly the world will be materially different in terms of emerging technologies and type of applications that they would enable and the type of capabilities we would required from the underlying networking and communication infrastructure.
This combined with the Internet architectural limitations and its ossification forces us to ask the questions -- should we be thinking about new paradigms, new Internet architectures, and services?
Given NSF’s mission and given research community interests, answer is of course yes…
New paradigms??
o different building block than datagram packet?
o information dissemination as the fundamental service and that should decide the fundamental building block -- maybe digital objects
o creating customized group communication networks on demand, that is, customized overlay networks on demand
o and maybe others
Around a new paradigm, we need to invent and design new network and protocol architectures and distributed systems and services infrastructure.
Please note that is what FIND Program is all about --
Clean-slate -- get the research community in a different mind set -- liberate the thinking from the constraints of Internet.
I want to mention that there are research teams that have already advocated clean-slate designs -- for example 100x100 project among CMU, Stanford, and Rice. Similarly NewArch project had also posed this question. And there are others.
Design and trial without roadblocks – clean-slate

13. NSF CISE Future Internet Research
NeTS
FIND: Future Internet Design
CyberTrust
Clean-slate secured network arch
CSR
New distributed system capabilities
CCF
Scientific Foundations for Internet Next Generation (SING)
CRI
Network infrastructure for arch research
MRI
Network instruments for arch research …
Experimental Facility to Validate Research
GENI

14. FIND
has much material, including full descriptions of projects
White paper program
Rolling requests for researchers with funding on architectural, big picture research
Clean-slate
Join collaborative program with other researchers
See the web site for more information on white papers

15. What is GENI?
GENI will be a full-scale programmable facility for research into the future internet
Focus: whatever experiments researchers need to create, test and evaluate the high impact new ideas
Clean-slate approach, Opportunity for:
Unencumbered starting points
Novel architectures and big picture
Strong coupling with technology drivers of futures
Mobile wireless
Novel optical substrate capability

16. GENI Design: Building Blocks
Three major components
Physical network ‘substrate’
expandable collection of building block components
federate
Software control & management framework
knits building blocks together into a coherent facility
embeds ‘slices’ in the physical substrate
Operational control (GENI management) is distinct from experiment control
Components
design, build, operate, iterate in modules, throughout lifecycle

17. GENI Design – Fully Programmable Routing/Switching Nodes
Clean-slate for architecture and protocol research
Slicing/Virtualization of Physical “Substrate”
Concurrent exploration of a broad range of experimental networks (including edge resources such as clusters)
Guaranteed resources, interference free will be available
Instrumented resources
Fully Measure experimental behavior
Interconnect GENI and the commodity internet
Users and applications can “opt-in” to GENI
There will be access to both “customers” for novel long-lived services and to “populations” of real users
Flexible and Phased Design
To support new technology introduction throughout GENI lifetime
See A Tale of Two GENI’s on GENI.net site!

18. Slicing and Virtualization
Sensor substrate
Fixed nodes
Mobile wireless substrate(s)
Conceptual framework to show how
o multiple network arhictectures can be supported on a single physical infrastructure
o user opt-in can work
To overcome the testbed dilemma

19. Federation Example Federated International Facility
Conceptual framework to show how
o multiple network arhictectures can be supported on a single physical infrastructure
o user opt-in can work
To overcome the testbed dilemma

20. Programmability Basic Approach
All network elements are programmable via open interfaces and downloadable user code
Uniform control and management architecture across all components – access node, core node, wireless…
GENI Control & Management Plane
API
Programmable
Edge Node
Programmable
Sensor Platform
Open API
Radio platform
Programmable core
Node

21. Wireless Substrate Considerations (presented to MobOpts RG July 25 2007)

22. NSF Wireless Testbed Experience
Programmable wireless and sensor network testbeds were developed in earlier NSF progams
ORBIT for evaluation of wireless network protocols
CitySense, Kansei open programmable sensor net testbeds
Coming this year: open underwater net
These open testbeds offer shared use and programmable layer 2/layer 3 protocols but full future internet impact needs full scale
ORBIT Radio Grid Emulator
ORBIT Field Trial Network
Harvard/BBN CitySense deployment plan

23. Candidate GENI Wireless: Dealing with Heterogeneity
This is the prototype design - a candidate platform
Single wireless GENI node architecture that covers different wireless device and network element needs:
Standard set of CPU platforms with different size/performance
Multiple radio cards as “plug-in” – easy to change radios, upgrade
Linux OS with appropriate “open API” drivers
GENI M&C
(“GMC”)
Linux OS
Linux OS w/
GENI control
Control Module
Plug-In radio modules
(evolving technology,..)
Processor Chassis with appropriate size/performance
(sensor GW, mobile node, ad hoc router, AP, BTS…)
End-user Wireless Devices
(commercial sensors, phones, PDA’s, laptops)
With GENI opt-in?
Wireless GENI Network Elements
All components also available as wireless kits

24. GENI Candidate: Programmable, Cognitive Radios
Several experimental programmable radio platforms under development for wireless network research…
WARP programmable radio, GNU radio, KU agile radio & near-future cognitive radios, ….
Key issue: open software API’s and protocol stacks for full control of physical and link/MAC layers
GNU Software Radio
KU Agile Radio
Lucent/WINLAB Cognitive Radio AP
Rice “WARP” board

25. Some Ongoing Discussions
How do we do systems engineering and planning for fast evolving radio / wireless substrates?
How much can radio / wireless be virtualized?
GENI performance in general?
How much scale-down is acceptable per element?
How can a project/experiment buy better performance fidelity, and how much?
Missing key radio (and optical, other physical) Areas?
Optical access networks?
Wireless over fiber?
Quantum Cryptography?

26. GENI Success Scenarios (many alternatives)
Internet evolution influenced by clean-slate approach
Architectures achieve fundamental progress in challenge areas such as security
New services and applications enabled
Alternate Internet architecture emerges
Virtualization becomes the norm with plurality of architectures
Single new architecture from the research enabled by GENI emerges and dominates
Alternative infrastructure becomes the mainstream over time
Many other payoffs, including unanticipated and high risk

27. Extra - Scope of GENI Optical Research?
Tight Integration of Physical Layer with Higher Layers
Provide Dynamic Optical Networking Plane
Dynamic Optical Switching/Routing
Fiber (Space) Switch (e.g. switch all WDM channels in a fiber)
Wavelength Switching (e.g. ROADM)
Sub-Wavelength Switching (circuit) - O/E/O
Optical Burst/Packet Switching?
All Optical Networking
Combines Transmission Issues with Optical Switching/Routing
Data Rates/Modulation Formats effect Network Design
Control Plane / Management
Optical Transmission?
Higher Data Rates, Longer Transmission Distance, Modulation Formats
Quantum Cryptography
Optical Device Technologies?

28. Optical Device Technologies?
Photonic Integrated Circuits (PICs)
Very dense and inexpensive OEO (scaling & cost reduction)
High Functionality PICs
Silicon photonics
Integration with electronics
Manufacturing Infrastructure Sharing (scaling & cost reduction)
Microstructured Optical Fibers
Customizable optical properties (transmission/all optical networks)
How would these breakthroughs change networking?
More efficient transmission or fundamental paradigm shift?
GOAL: Flexible GENI design to allow Introduction of New Technologies as they mature