Download IEEE FIPA P2P Nomadic Agent Kick off presentation

P2P Nomadic Agent WG
Scope:
P2P Nomadic Agents are autonomous agents capable
of running on nomadic devices in a P2P world, without
need for a connection to a server or a container.
Chair: Bernard Burg, Panasonic, US
Co-Chair: Heikki Helin, TeliaSonera, FI
Co-Chair: Michael Berger, Siemens, DE
s
Sept. 13 2005, Budapest, HU
© IEEE FIPA P2P Nomadic Agent WG
Problem Statement
Current
peer-to-peer (P2P) standards (UPnP Forum
www.upnp.org) need an additional level to deploy and
manage applications automatically. Agent technology is
uniquely well-suited for this task, because it is
intrinsically P2P capable of managing configurations and
leading negotiations in serverless environments (also
called pure P2P in the following).
FIPA
is the right venue for this effort
 FIPA
has already produced a standard on Ad-hoc environments
which can be leveraged.
 FIPA is a fast moving lightweight standard and time is of the
essence.
Sept. 13 2005, Budapest, HU
2
© IEEE FIPA P2P Nomadic Agent WG
New Motivation
 Potential
impact: P2P networks represent an unparalleled growth
opportunity for the agent community to push FIPA specifications into
pervasive products ranging from telephony to consumer electronics.
 Potential
 The
application fields for P2P Nomadic Agents: Huge
P2P market is growing fast;
 Skype
has 47 million users and adds 150,000 new ones every day, with
an average 2.5 million Skype users online at any time.
(source www.skype.com)
 In total, P2P network traffic may account for up to 85% of Internet
bandwidth usage, mostly for file transfer applications like bitTorrent,
eDonkey2000, Kazaa …
(source http://linuxreviews.org/news/2004/11/05_p2p/).
Sept. 13 2005, Budapest, HU
3
© IEEE FIPA P2P Nomadic Agent WG
A P2P Nomadic Agent Use Case: Home Environment
Our life starts to be invaded by P2P
nomadic devices
We love it! because it:
 provides a superior experience
thanks to specialized devices in a
competitive market
 allows (free) P2P services
 updates automatically
 does not require computer literacy
 gives an integrated environment
IP Camera
Mobile Phone
Skype Phone
However, to do this, there is a need of:
 an intelligence on top of the existing
P2P systems, Agents are the best
solution because of their distributed
reasoning and negotiation
capabilities.
 a P2P Nomadic Agent Standard
Sept. 13 2005, Budapest, HU
4
Computer
HelloKittyROBO
Plasma TV
Rice Cooker
Answering system
Home Entertainment System Alarm system
These devices are network capable through WiFi, bluetooth, PLC …
© IEEE FIPA P2P Nomadic Agent WG
Objective
The
objective is:
 to
define a specification for P2P Nomadic Agents, capable of
running on small or embedded devices, and
 to support distributed implementation of applications for
consumer devices, cellular communications and robots, etc. over
a pure P2P network.
This
specification will:
 leverage
presence and search mechanisms of underlying P2P
infrastructures such as JXTA, Chord, Bluetooth, etc.
 propose the minimal required modifications of existing FIPA
specifications to extend their reach to P2P Nomadic Agents.
Sept. 13 2005, Budapest, HU
5
© IEEE FIPA P2P Nomadic Agent WG
Forward to the Past
Sept. 13 2005, Budapest, HU
6
© IEEE FIPA P2P Nomadic Agent WG
TC Ad-hoc
Scope: Enhance interoperability between FIPA agents
operating in ad-hoc environments
Chair: Michael Berger, Siemens
Co-Chair: Heikki Helin, Sonera
Editor: Michael Pirker, Profactor
Sept. 13 2005, Budapest, HU
© IEEE FIPA P2P Nomadic Agent WG
Problem Statement
Mobile
computing
 GPRS,
GSM, small devices
 FIPA has several solutions, e.g. bit-efficient ACL and envelope
encoding for low bandwidth connections
 Used in several projects (e.g. LEAP, CRUMPET)
Mobile
Ad-hoc Computing -> FIPA needs a solution
 New
technologies for short range wireless communication (e.g.
WLAN, Bluetooth) and Pervasive Computing (sensing/acting in
embedded systems)
 Devices come into communication range ->
-> Agents on two mobile devices, originally created on different
platforms, need to dynamically discover each other and share
services. Limited resources of devices have to be considered.
Sept. 13 2005, Budapest, HU
8
© IEEE FIPA P2P Nomadic Agent WG
Problem Statement
Current
Technologies
 Dynamic
message routing over multi-hops on network layer
(MANET protocols)
 Different
technologies to describe, discover and share services:
as API to infrastructure (e.g. JINI, Salutation) or as protocol
specification (e.g. UPnP, SDP)
 Dynamic
service discovery for fixed (Internet-based) P2Pnetworks (SLP, JXTA, Gnutella) may also be used because of the
same nature -> large variety of approaches (with central
elements, pure P2P solutions, advanced solutions with intelligent
distribution / replication of service directory entries; mostly
application oriented)
Sept. 13 2005, Budapest, HU
9
© IEEE FIPA P2P Nomadic Agent WG
Focus
Use
of existing approaches which provide support on
different levels of discovery, e.g.:
 Bluetooth
and WLAN ad-hoc networks,
 Plug and Play mechanisms like Jini and UPnP,
 P2P approaches like JXTA
Specifically
developed for dynamic environments and
service discovery -> more efficient search
Adhoc
/ P2P mechanisms also useful for fixed networks
Sept. 13 2005, Budapest, HU
10
© IEEE FIPA P2P Nomadic Agent WG
Sept. 13 2005, Budapest, HU
11
© IEEE FIPA P2P Nomadic Agent WG
New specification 00095
Description
Approved
JXTA DM (according to Annex C in 00023)
as preliminary Specification PC00095A
Implemented
by Sergi Robles and team at Universitat
Autonoma de Barcelona, SP.
Sept. 13 2005, Budapest, HU
12
© IEEE FIPA P2P Nomadic Agent WG
Back to the Future
Sept. 13 2005, Budapest, HU
13
© IEEE FIPA P2P Nomadic Agent WG
Technology
 Extend
existing FIPA spec PC00095A towards general support of P2P,
allowing leveraging of presence mechanisms, directories and search
capabilities and linking to the existing FIPA search mechanisms in a
seamless manner
 Define a lower-level API connecting this FIPA P2P Nomadic Agents to P2P
networks
 Use the FIPA ACL
 Enable ad-hoc connections to small networks of peers (PAN, LAN…) and
across these networks
 Allow the self-organization and management of a set of devices in a pure
P2P network
 Ensure the ability to run on a small footprint platform or an embedded
device (consumer electronics, nomadic devices, cellular phones, etc.)
 Optional:
 Extend
the FIPA search mechanism towards more generic searches.
Provide a layered approach to semantic searches
 Define a higher-level API linking this P2P Nomadic Agents to existing FIPAcompliant tools
Sept. 13 2005, Budapest, HU
14
© IEEE FIPA P2P Nomadic Agent WG
Outside of our scope
ad-hoc
connections to the whole Internet or very large
scales. Our focus is on small networks (<100 nodes tbd)
mobile
code and mobile agents.
 The
defining notion of 'mobile agents' is distinct from, and should
be distinguished from, the notion of 'nomadic agents'. It is the
latter concept, 'nomadic agents', that is definitive to the central
concerns of the working group (Martin Purvis)
 Hence our WG will not preclude mobile agents, but will not
specify their problematics. Our focus is on P2P Nomadic Agents.
Sept. 13 2005, Budapest, HU
15
© IEEE FIPA P2P Nomadic Agent WG
Workplan, Milestones, and Deliverables
June 2005: FIPA joins IEEE
97
Experimental
specs
96
creation
98
Experimental
specs
FIPA
2000
2002
Approved
Specs
P2PNA
P2P Nomadic Agents
28 Aug
WG creation
29 Jul
FIPA meeting
Utrecht
Reference
Implementation
30 Sept
Oct 06
CFP
13-14 Sept
Dec 05 CFP
March 06
June 06
Sept 06
Dec 06
Kick-off
meeting
Answers,
Technical
Directions
Draft
Preliminary
Specs
1st Reference
Implementation
Experimental
Specs
Approved
Specs
Most of the work will be done by email and conference calls in between
face-to-face meetings. Meetings are devoted to decision making and
solving hard issues that cannot be solved by any other means.
Sept. 13 2005, Budapest, HU
16
© IEEE FIPA P2P Nomadic Agent WG
Submitters
 Michael
Berger (Siemens, DE),
 Bernard Burg (Panasonic, US),
 Heikki Helin (TeliaSonera, FI).
Michael, Bernard and Heikki have significant experience, each of them
having been FIPA specs editors and TC chairs where we led efforts
on mobility, management, naming, message transport, and ad-hoc
specifications that helped push FIPA towards new horizons.
We have also led projects outside of FIPA to implement Agent-related
platforms in LEAP, Micro-FIPA OS; worked on ad-hoc projects on
JXTA and Bluetooth; and have been involved in Agentcities aimed at
spreading FIPA agent technology at large.
Sept. 13 2005, Budapest, HU
17
© IEEE FIPA P2P Nomadic Agent WG
Submitters
We are committed to the success of this enterprise and
have the good fortune to assemble a good mix of
enthusiastic supporters from both academia and industry
from around the world. As a group, we are capable of
driving the vision as well as pushing it towards products.
Sept. 13 2005, Budapest, HU
18
© IEEE FIPA P2P Nomadic Agent WG
Supporters (confirmed on Sept 7 05)
 Stephen
Cranefield (U. Otago, NZ),
 Itsuki Noda (AIST, JP),
 Yoshiharu Dewa (Sony, JP),
 Michal Pechoucek (Gerstner Labs, CZ),
 Tim Finin (UMBC, US),
 Michael Pirker (Siemens, DE),
 Martin Griss (CMU-West, US),
 Martin Purvis (U. Otago, NZ),
 Oleg Karsayev (SPIIRAS, RU),
 Sergi Robles (UAB, SP),
 Alberto Y.B. Kim (Samsung, KR),
 John Shepherdson (BT, UK),
 Matthias Klusch (DFKI, DE),
 Carles Sierra (IIIA, SP),
 Ryszard Kowalczyk (Swinburne U. AU),  Santtu Toivonen (VTT, FI),
 Alan Messer (Samsung, US),
 Arkady Zaslavsky (Monash U., AU).
 Philippe Morin (Panasonic, US),

 EU
IST CASCOM project (www.istcascom.org)
Sept. 13 2005, Budapest, HU
19
© IEEE FIPA P2P Nomadic Agent WG
Outline of Budapest meeting
Organization
of WG life
 Logistics
 Meetings
 Mailing
list (where, who, additional members?)
Refinement
of Objectives
 Workplan
update
 Work on comments
Edition
of the Call for Comments
You are invited to join this exciting adventure!!
Sept. 13 2005, Budapest, HU
20
© IEEE FIPA P2P Nomadic Agent WG