PeertoPeer activity at PRL

1
Peer-to-Peer activity at PRL
2
Outline

• Introduction Opportunities
• Current projects
• VoD on Residential Home Gateways (customers
  ISPs)
• VoD Live streaming on PCs (customers
  e-tailers)
• Where we are where we go

3
The P2P revolution

• File sharing (Napster, Gnutella, KaZaA, eDonkey,
  BitTorrent,)
• Telephony (Skype,)
• Live Streaming (SopCast, PPLive, Octoshape,)
• VoD (AOLs IN2TV, Skys Sky by Broadband,)
• BitTorrents agreements with TV channels and
  studios
• Venice project by KaZaA founders

4
What P2P is about

• NOT just illegal file sharing
• Scalable resource pooling

capacity
5
Business impact

• Telcos / ISPs
• Little control over P2P traffic routes
• b/w commoditization (i.e. no profit if no
  service)
• Content creators
• Cheap P2P distribution enables emergence of new
  content creators
• Broadcasters and e-tailers
• P2P a global distribution channel

6
Electronic content delivery landscape
End users
ISPs
Content aggregators / E-tailers
Content creators
7
Opportunities the e-tailer segment

• Key players in the content delivery chain
• In search of ISP-independent delivery channel
• P2P system of end-users PCs meets their needs
• Thomson launched project to develop such system
• Initial target VoD service

8
Opportunities the ISP segment

• Residential home gateways (RHG)
• A unique location for P2P service deployment

• Controlled environment, with predictable
  behaviour
• Managed by ISP
• Software hard to hack
• Always on
• Good for
• Service availability
• Content security

9
Opportunities the ISP segment (ctd)

• P2P services on RHGs
• VoD general community-specific content
• Live Streaming TV channels Podcast
• Gaming
• P2P removes game server bottleneck
• Cheating prevented in controlled environment
• Network troubleshooting
• See Augustin Soules talk

10
Other potential targets

• Controlled environments
• In-Flight Entertainmnent Air France, Thales
• On-the-road Entertainment (taxis, trains)
• Hotel rooms
• Ad hoc networks (phones, PDAs,)
• Applications
• VoD, Live streaming, Gaming
• Location-sensitive content search (point me to
  nearest dentist)
• See Augustin Chaintreaus talk

11
Why us

• Key expertise
• STBs RHGs
• Content security
• Network troubleshooting Paris Lab
• P2P technology Rennes, Princeton, Paris Labs
• Privileged relations
• Hollywood studios
• France Telecom, Telefonica

12
Why now

• Today
• Commercial offers on PC platform (Sky, AOL,
  Venice Project, BitTorrent)
• OS support (Windows Vistas P2P SDK)
• Tomorrow
• ISPs
• All e-tailers and content aggregators

13
Outline

• Introduction Opportunities
• Current projects
• VoD on Residential Home Gateways (customers
  ISPs)
• VoD Live streaming on PCs (customers
  e-tailers)
• Where we are where we go

14
Push-to-peer VoD system overview
Control server
Video server
RHGs
. . .
DSLAM
15
Push-to-peer push-phase
Control server
Video content server
RHGs
. . .
DSLAM
16
Push-to-peer The pull-phase
Control server
Video content server
RHGs
. . .
DSLAM
17
Why Push-to-peer?

• No ISP bandwidth consumption beyond DSLAM
• Retains advantages of content server-based
  solution
• Guaranteed content security
• Short playback delays
• But at a lower cost
• More robust (content server single point of
  failure)
• No need to overprovision content server uplink
  b/w

18
Design challenges

• Ensure efficient resource pooling
• System equivalent to single content server with
• Storage sum of individual storage spaces
• B/w sum of individual uplink bandwidths

?
. . .
19
Solution (Push-phase) data format
File prefix
Data window

• Proposed solution
• Prefixes to reduce startup latency
• Encoding to increase flexibility (can recover
  content from any sufficiently large set of peers)

Window prefix
Encoded data
20
Solution (Pull-phase)

• Pull data from least loaded boxes
• Provides load balancing
• Load information available at control server

RHGs
21
Dimensioning analysis

• Models to predict startup delays for varying
• Rate of movie requests
• Memory

22
Push-to-Peer current status

• Dimensioning analysis used to
• Optimise push phase w.r.t. movie popularity
• Identify system requirements (memory and
  bandwidth) for given demands
• Ongoing work
• Wireless prototype demonstrator of IFE for Air
  France

23
Outline

• Introduction Opportunities
• Current projects
• VoD on Residential Home Gateways (customers
  ISPs)
• VoD Live streaming on PCs (customers
  e-tailers)
• Where we are where we go

24
Live Streaming project overview (ctd)

• Key components
• Communication graph construction
• Scheduling encoding
• (what to exchange with whom)
• Infrastructure / relay boxes

25
Intra-flow scheduling

• Main challenge manage tension between
• Timeliness
• Diversity
• Proposed packet forwarding strategy
• Random Useful packet forwarding
• ? Proved to use b/w resources optimally
• for ANY communication graph

a
b
c
26
Inter-flow scheduling

• Aim Minimize playback latencies among flows
• Means allocate uplink bandwidth optimally to
  competing flows
• ? Open research problem!

27
Network coding

• Can reduce playback delay and increase streaming
  rate
• Simplifies signaling and scheduling
• Opportunity develop dedicated chip
• ? allows better codes at affordable
  encoding/decoding speeds

28
Infrastructure boxes

• Could be deployed in ISP networks, to
• Speed up live streams delivery
• Reduce ISP bandwidth consumption
• Can host coding chip
• Potentially makes STB/RHG streaming service more
  widely deployable
• Open question to meet target Quality of Service
• How many boxes?
• Deployed where?

29
Outline

• Introduction Opportunities
• Current projects
• VoD
• Live streaming
• Where we are where we go

30
Achievements

• 3 papers accepted in IEEE Infocom 2007 conference
• 2 papers submitted to ACM Sigmetrics 2007
• One Eurecom Master Thesis (Matteo Varvello)
• 11 patents filed, in collaboration with Rennes

31
Collaborations

• Within Thomson
• Corporate Research (Rennes and Princeton)
• STS
• With Academia
• INRIA/IRISA
• Eurecom
• Princeton University
• University of Massachusetts
• Industrial partnerships
• Potential European Project within FP7 with
  Telefonica

32
Conclusions

• All future large-scale communication content
  distribution systems will rely on P2P
• Thomson can play a role in this space
• P2P key activity of the lab, bridging all our
  projects
• Network anomaly detection troubleshooting
• Wireless
• Ad hoc networking