InBCT 3.2 Peer-to-Peer communication/computation Cheese Factory -project http://tisu.it.jyu.fi/cheesefactory

1
InBCT 3.2Peer-to-Peer communication/computationC
heese Factory -projecthttp//tisu.it.jyu.fi/chees
efactory

• Prof., D.Sc. (EE) Jarkko Vuori
• Faculty of Information TechnologyDept. of
  Mathematical Information TechnologyEmbedded
  Systems
• Jarkko.Vuori_at_jyu.fihttp//tisu.it.jyu.fi/embedded
  /http//tisu.it.jyu.fi/mimic/

2
Agenda

• What is P2P (shortly)?
• InBCT 3.2 research project
• 2003Q3 status
• 2004 (in future)

3
P2P Introduction

• Recently, the peer-to-peer (P2P) paradigm for
  building distributed applications has gained
  attention from both industry and the media
• Peer-to-peer classical definition
• A P2P system is composed of a distributed
  collection of peer nodes
• Each node is both a server and a client
• may provide services to other peers
• may consume services from other peers
• Completely different from the client-server
  model, where
• Few specialized servers provide services to a
  large number of clients

4
Why P2P?

• Despite its poor reputation, P2P is extremely
  interesting from a technical point of view
• Its completely decentralized model enables the
  development of applications with
• high-availability
• fault-tolerance
• scalability
• characteristics previously unseen in Internet
• It exploits what has been defined the dark
  matter of Internet
• Unused CPU and storage capacity (in idling PCs)
• Moreover, P2P is not limited to file-sharing, but
  it can be applied to distributed computing and
  collaboration tools
• P2P can also be viewed as a robust information
  diffusion media

5
P2P Services

• Areas of applicability of P2P
• sharing of information
• information diffusion over net
• distributed web servers, distributed media
  repository
• sharing of storage capacity
• distributed file system, distributed search
  engine
• sharing of CPU time
• parallel computing
• sharing of human presence
• the P in P2P is Person
• sharing of connectivity
• resilient overlay networks

6
Peer-to-Peer Key Questions

• Does it work?
• can we find the data? (main InBCT 3.2 topic)
• Because data communication is already discussed
  under the Ad Hoc -topic
• query success rates
• length of query paths
• Does it scale?
• logarithmic / linear / polynomial
• Is it robust?
• participants are unreliable (future InBCT
  research topic)
• different failure modes possible
• Is it efficient?
• using communication resources efficiently
• Information diffusion
• the speed of information spreading (future InBCT
  research topic)

7
P2P Mobileencounter networks

• Information distributes over mobile device
  encounters (Mobile P2P is a future distribution
  model)
• no centralized server, zero configurability, free
  communication bandwidth, no infrastructure
• Applications
• information distribution
• e.g. cheapest bulk product search (gasoline)
• gasoline payment with mobile device
• mobile devices communicates with each other (e.g.
  Bluetooth)
• everybody tells what he/she has paid for the gas
  and gets in exchange prices of other gas stations
• based on this information, mobile device can
  recommend the cheapest place to fill the tank
• boosts the market based economy by giving equal
  info over the market situation to all participants

8
Why to make research in this field

• Scientific (and education) aspect (universities)
• resource finding problems
• modelling information diffusion
• Social aspect (universities and companies)
• changes the structure of the society (and social
  code)
• Mobile terminal builder
• this new information diffusion method enables new
  applications
• Games
• Software maker
• How to implement those P2P-systems
• Operator
• P2P distributed media will be opened by
  certificate
• certificate is moved over operators network
• Content producer
• new distribution channels needs new content

9
Structure of the project
InBCT
Yomi
Department of MIT
? Power-law basic research
Mikko Vapa Joni Töyrylä Adaptive
search algorithms (neural computing)
Sergiy Nazarko Distributed data fusion
Oleksiy Volovikov Gasoline Price Comparison Syste
m
Niko Kotilainen Joni Töyrylä P2P
network simulator
Niko Kotilainen Peer-to-Peer Studio
Annemari Auvinen Management of connections
and topology
MoPeDi Mobile Chedar / BlueCheese
Matthew Kam Information diffusion simulator
Niko Kotilainen Chedar P2P platform
Tommi Myöhänen P2PCore Search algorithms
Jani Kurhinen Vesa Korhonen Diffusion
of information in P2P networks
Matthieu Weber Formation of P2P networks
Vagan Terziyan Hermanni Hyytiälä Eetu
Ojanen Semantic web in P2P networks
Sauli Takkinen New applications of mobile
P2P networks
Theoretical Practical
10
Status of the project

• P2P platform Chedar is working
• Distributed data fusion
• running ok on the Chedar platform
• simple test set-up now, full demo version ready
  at Q4
• Search algorithm (resource discovery in P2P
  networks)
• Neural optimized algorithm working ok
• Report will be ready Q2, publication Q2, 2nd
  publication Q4
• Formation of P2P networks
• First simulator version running now
• Publication ready Q2
• Masters Thesis
• Annemari Auvinen (Chedar and its algorithms) will
  be ready Q4
• Joni Töyrylä (analysis of NeuroSearch) will be
  ready 2004
• Tommi Myöhänen (Search algorithm comparison) will
  be ready 2004
• Oleksiy Volovikov (Gasoline price comparison
  system) will be ready 2004
• Licentiates thesis
• Jani Kurhinen (information diffusion modelling in
  mobile P2P environment) will be ready 2004

11
Next steps (2004 Q1)

• Distributed datafusion application development
• Full distributed datafusion (fault tolerant) demo
  version
• Neuroprogrammed search algorithm
• Why this algorithm is working so well?
• How to apply to mobile environment (where
  topology is changing rapidly)?
• Development of the P2P-network simulator
• NS2 does not support training of neural networks
• Modelling/Simulating information diffusion in P2P
  environment
• First simple simulator running
• Development of the first version of mobile-P2P
  platform using Bluetooth