Welcome

• 31 submissions
• 9 papers
• 7 posters

Keynote

Mobile P2P evolution from ad-hoc routing to indexing and content sharing: new models, opportunities, challenges.

Mario Gerla

• P2P on wired internet started as app overlays
  • then Distributed directories
  • then distributed content
• "Pull the chicken by the neck" model for P2P
• overlay:
  • RON
  • End-system multicast
• file sharing
  • Chord overlay

P2P in wireless

• Early efforts were ad-hoc/P2P (1973)
• IETF MANET: AODV/DSR/OLSR
  • Only partial success
    • Not optimal for all configs
    • usually used just as benchmarks

Ad-Hoc customers:

• Military:
  • Automated battlefield
• Civilian:
  • Disaster recovery
  • Law enforcement
  • Homeland dfense
  • Search and rescure
  • Space/planet explanation
• Commercial:
  • Vehicle nav safety
  • Mesh network
• Ignoring WSN because it's separate thread
• Challenges in tactical networks
  • Scalable routing
  • many nodes
  • dynamic topo
  • QoS
  • hostile environemnt
• "Landmark" routing
  • group-based, assumes units moving togethr.
  • landmarks elected.
  • Landmarks advertise position to whole net.
  • packets don't have to go through landmark
  • still suffers from multi-hop poor TCP/QoS performance
  • UAVs to shorten paths

Paradigm shift:

Move from accessing available Internet to "Bypass Internet" to get new functionalities

• Opportunistic Mesh
• Vehicle Mesh
• Opportunistic Piggyback transmitting file block by block on passing vehicle
• DSRC: 802.11p Vehicle Grid Enabler

Belief: opportunistic ad-hoc networking paves the way to P2P networking

• Vehicle P2P examples:
  • CAR Torrent
  • Mobeyes

Mobeyes: distributed data collection for event reconstruction

UCLA: 50 vehicle campus testbed U-veT

Concluding observation:

1. cannot "transfer" existing ad hoc protocols directly to mobile P2P apps
2. mobility: must study correlated mobility, etc
3. peer collaboration: cooperation, incentive, protection
4. interaction: location aware, pervasive, filtering/fusing
5. wired-wireless p2p: TCP issues on multihop, exploit wired net, congestion
6. storage/query processing: mobile DB: how to mine and filter

Future:

• Understanding mobility
• Robust to channel
• Cross-layer opp

Bad news

Mobile P2P can not blindly inherit previous ad-hoc research

Good news

new exciting research direction, appealing applications that may become widespread

MobiShare Session 1

MobiShare Poster Lunch?

MobiShare Session 2

MobiShare Session 3?