From the Rocks of Mars to the Slums of Cambodia: The ...msakr/15129-f10/Docs/K... · From the Rocks...
Transcript of From the Rocks of Mars to the Slums of Cambodia: The ...msakr/15129-f10/Docs/K... · From the Rocks...
From the Rocks of Mars to the Slums of Cambodia: The Unexpected Evolution of Research
Khaled A. Harras, Ph.D.
Assistant Teaching Professor Founder and Director of the Networking Systems Lab (NSL)
School of Computer ScienceCarnegie Mellon University
Outline
• Story 1: – Evolution: From 4 nodes to 1.1Billion
• Story 2: – Operation: How does all this roughly work?
• Story 3: – Mobility: Common problems and solutions
• Story 4:– DTNs: The unexpected evolution of research
The Internet in 1969…
Source:http://www.sharpnet.co.uk
The Internet in 1971…
The Internet in 1977…
The Internet in 1999…
Source:http://www.caida.org
The Internet: Some Stats…
…More Stats…
…More Stats…
Source:http://broadbandanalyst.co.uk/
Asia has “started” using the Internet!
Story Punch-line…From To
• Research is the core for enabling this…
Research Results…
Outline
• Story 1: – Evolution: From 4 nodes to 1.1Billion
• Story 2: – Operation: How does all this roughly work?
• Story 3: – Mobility: Common problem and solutions
• Story 4:– DTNs: The unexpected evolution of research
Class Experiment…
• Small game to send messages…• What are the problems we can have?
– OPEN DISCUSSION…
Protocol Layers
Source: http://www.fidis.net/resources/deliverables/hightechid/int-d37003/doc/6/ and http://www.satmagazine.com/cgi-bin/display_article.cgi?number=1342678280
Networking Problems
• What are the problems we can have?– Connection establishment– Reliability– Flow control– Error control– Multiple applications– Many protocols– Routing– Congestion Control– …….!
Example: Routing
Source: http://www.h3c.com/portal/ and http://www.trainsignaltraining.com/
• Build Routing Tables
• Lookup address• Forward packets
Outline
• Story 1: – Evolution: From 4 nodes to 1.1Billion
• Story 2: – Operation: How does all this roughly work?
• Story 3: – Mobility: Common problems and solutions
• Story 4:– DTNs: The unexpected evolution of research
Today’s Internet: Simple View
Can we handle mobility… ?
• Increased wireless network deployment, and available Mobile devices
• High mobility and user expectation: “I must have network access…
NO MATTER WHAT..!!!”
Mobility: Just the beginning…
Source: Chetan Sharma Consulting
Solutions: Single-hop wireless
Solutions: Ad hoc Networks
Solutions: Mesh Networks
So… Can we handle Mobility..?
Outline
• Story 1: – Evolution: From 4 nodes to 1.1Billion
• Story 2: – Operation: How does all this roughly work?
• Story 3: – Mobility: Common problems and solutions
• Story 4:– DTNs: The unexpected evolution of research
Today’s Internet: Unstated Assumptions
• A path exists between endpoints– Routing protocols find the best path,
or even “a path”• Small end-to-end RTT
– Millisecond range• End to end reliability works well
– Especially for low data loss rates– Loss = Congestion
• Packet switching is the “right” abstraction– IP does best effort delivery for each packet
separately
The Rocks of Mars…
• In 1998 the CCSDS (Consultative Committee for Space Data Systems) founders teamed with Vint Cerf to develop the IPNSIG formed in September 1999.
• Objective: Define the architecture and protocols for an InterPlanetary InterNet: IPN
• Vision: A network of Internets
Inter-Planetary Internet A 30 year vision
Fine, those guys are just crazy… Why do “we” care..?
The Rise of New Applications
New “Common” Assumptions• Very large delays
– RTT could range from milliseconds to days– Ex: propagation between Earth and Mars
• Intermittent/disconnected links and paths– End to end connectivity might “never” exist– Routers may need significant persistent storage
• High link error rates– RF interference, light or acoustic interference– Loss is due to corruptions, not congestion
• Heterogeneous underlying networks and protocols– Some specialized networks may not run IP, and run
customized protocols that better suit them
IPN “Upgraded” to DTN Delay/Disruption Tolerant Networks
• DTNRG was formed in 2002 to generalize concepts introduced by the IPNRG
• Goal: achieve interoperability among heterogeneous networks in “extreme environments”
• Outcome: DTN Architecture and many protocols
How about other “simpler” applications…?
The Slums of Cambodia…
Are you kidding..! Who will pay for this…?
4 Billion PeopleEarning less than $2,000/year
< $2,000 4,000
‘The Bottom of the Pyramid’
Middle Classin developing
countries$2,000—$20,000 2,000
WealthyNations
Annual Purchasing Power in $US
> $20,000
PopulationIn Millions
100
Source: Prahalad & Hammond, Harvard Business Review, Vol. 80, Issue 9 (Sep. 2002), pp48-58
emerging‘mass’ markets
adjacent markets
The Bottom of the Pyramid• 3-4 billion people with per-capita equivalent
purchasing power less that US$2,000 per year• Could grow to 6-8 billion over the next 25 years• Most live in rural villages or urban slums and shanty
towns• Motivated towards urbanization and education• Dharavi, one of the poorest villages in India:
– 85% have a TV, 50% have a pressure cooker, 21% have a telephone… but can’t afford a house!
• Even the poorest of the poor in Bangladesh:– Devote 7 percent of income to communications
services (GrameenPhone)
http://www.wri.org/meb/wrisummit/pdfs/hart.pdf
The future of DTNs
• Mobile devices are starting to converge and have access to multiple networks
• Shouldn’t we take advantage of these multiple networks? Specially in challenged environments?
• Numerous Problems...
Outline
• Story 1: – Evolution: From 4 nodes to 1.1Billion
• Story 2: – Operation: How does all this roughly work?
• Story 3: – Mobility: Common problem and solutions
• Story 4:– DTNs: The unexpected evolution of research
• What to take from this talk…?
Research is a MUST!
Thank You..!
Questions…?
Contribution I: Delay Tolerant Mobile Networks (DTMNs)
• DTMN Architecture– Controlled flooding schemes– Evaluation via simulation
• Summary and Conclusions– We can significantly reduce the cost of
flooding– Few high-end nodes make a big
difference– The challenge is to “tweak the knobs”
D
S• MANETs rely on end-to-end routes• What do we do in sparse mobile
networks?– At that point in time: Flooding
delivers… too expensive– Can we “simply” reduce the cost?
Contribution II: Clustered DTMNs and Messenger Scheduling
• Summary and Conclusions– No “best scheme”; tradeoffs depend
on the network/traffic patterns– Picking the “best” scheme is a
challenge and depends on which metric has more importance; we provide such knowledge
– Adaptive strategies are very useful in such scenarios D
S
• What happens if “clusters” are formed?• Clustered DTMNs
– Use dedicated “messengers” between “regions”
– Point to point vs Path travel– Messenger management: Ownership
and Scheduling Algorithms– Evaluation via analysis and simulation
Contribution IV: Intermittent Connectivity
• Summary and Conclusions– Improvement due to bundling, single
connections, and smaller RTT– Mini-bundles is a major parameter– Solution requires no network change for
deployment
D
S• Shift focus to a single device experiencing “Disruption” rather than “Delay”
• Intermittent Connectivity & DBS-IC– View data while offline– An SA “bundles” data, delivers to MA
whenever possible– Evaluation via implementation
AP
AP
AP
• How do we make use of available opportunistic connections?
DBS-IC
ParaNets