Mobile Agents

1
Mobile Agents

• Martin Beer,
• School of Computing Management Sciences,
• Sheffield Hallam University, Sheffield,
• United Kingdom
• m.beer_at_shu.ac.uk

2
Overview

• Introduction to Mobile Agents
• What are they?
• How can they be used?
• A Currentb Example
• The LEAP Project
• The Future

3
Mobile Agents

• software processes which can
• roam wide area networks
• interact with foreign hosts
• perform tasks on behalf of their owners
• return home

4
Mobile AgentsKey Hypothesis

• in certain applications, they provide practical,
  though non-functional, advantages which escape
  their static counterparts
• imagine having to download many images just to
  pick out one. Is it not more natural to get your
  agent to go to that location, do a local search
  and only transfer the chosen compressed image
  back across the network?

5
Part View of Telescript Architecture
6
Telescript Concepts
7
The Telescript go operation

• Host engine
• requires destination place
• packages up the agent along with all its data,
  stack and instruction pointer
• ships it off to destination place

• Destination engine
• unpacks agent
• checks its authentication
• agent now free to resume execution at this new
  place
• on completion of tasks, agent returns to original
  host

8
Mobile AgentsDevelopment Languages

• Telescript
• Java
• Agent-Tcl
• Safe-Tcl
• Xlips
• C/C
• indeed any programming language

9
Mobile AgentsApplications
10
Mobile AgentsKey Challenges

• transportation
• authentication
• secrecy and privacy
• destination system security
• cash
• performance issues
• interoperability/communication/brokering services

11
The LEAP Project

• Acronym Lightweight Extensible Agent Platform
• Reference IST-1999-10211
• Effort 433.6 Man/Months
• Budget 5.4 Millions Euros
• Run Jan 2000 - June 2002
• Consortium
• Motorola (prime contractor)
• ADAC
• British Telecommunications
• Broadcom
• Telecom Italia Labs
• University of Parma
• Siemens

12
The mobile workforce

• Commercial, maintenance, assistance activities
  are inherently mobile.
• A trend which has been revolutionized by the new
  methods of communication (cellular phones and
  Internet), and which extends to new kinds of
  mobile workers.
• Remaining issues
• Centralized management is not suited to this
  type of architecture
• No/limited access to information
• Expertise/Knowledge of the mobile team is under
  exploited
• Lack of social integration.

13
The LEAP objectives
Develop an architecture and applications, which
support a mobile enterprise workforce.

• Decentralised teamwork co-ordination
• Collective decision making
• Flexibility in work scheduling (e.g. job
  swapping)
• Creation of virtual teams
• Travel management
• Anticipating workers travel needs
• Providing guidance and time estimation
• Synchronising the movements of virtual teams
• Knowledge Management
• Anticipating workers knowledge requirements
• Accessing and customising knowledge
• Networking individuals with each other, based on
  their expertise

14
LEAP applications

• Agent an autonomous piece of software that acts
  on behalf of its user and communicates,
  co-operates and negotiates with its peers in
  order to achieve its goals.
• Agents and the mobile workforce
• Some similarities (distributed, autonomous, goal
  driven)
• Dynamic and open architecture (flexible, peer to
  peer)
• Customize services according to users specific
  needs
• Enable emergence of intelligent behaviours.
• Generic services swap-shift, plan-route,
  make-collective-decision, find-expert
• JADE-LEAP a standard agent platform that runs on
  devices from phones to PDA to desktops.

15
Technical scenario
Jobs DB
Specific work request
Selection of Field Engineer based on personal
preferences
Work Scheduler
Customer
Repair fault Information (Expert,
Technical) Update (Experience)
Geogr. DB
Travel to Job Guidance (Route planning) Informatio
n (Traffic)
Field Engineer
Find-relevant-information
Find-expert
Plan-route
Traffic
Expertise DB
Update-knowledge
Find-relevant-information
Estimate-route-cost
16
Social scenario
Social Events DB
Geogr. DB
Jobs DB
Field Engineer
Field Engineer
Trade work (Shifts, Holidays, Overtime) Organize
meetings (Event, Location, Time)
Field Engineer
Plan-route
Swap-shift
Field Engineer
Make-collective-decision
Coordinate-social-activity
17
Field trials
ADAC - Yellow Angel Trial ADAC employs 1,700
Yellow Angels for road-side assistance in
Germany performing 9600 incidents / repair cases
per day while driving 150,000 km. Time
March-April 2002, duration 10 days Localition
area of 100 km around Munich, Germany
Participants 5 yellow angels (in shifts, more
than 2 per shift) How Emulate several car
breakdowns per day BT - Field Engineer Trial BT
has 25,000 engineers performing 150,000
installation and repair tasks each day in the
UK. Time March-April 2002, duration 14
days Localition area of Birmingham,
UK Participants 10 Customer Service Team members
(more than 2 per field unit to have some social
interaction) How Parallel working with current
and LEAP systems within real work situations
18
What was Achieved?

• Achievements
• A FIPA compliant agent platform that runs on all
  Java editions (J2SE, J2ME, pJava, J1.1.x).
  Released in open-source on September 26th, 2001
• Validation of concept on a laboratory trial
  application the cinema-organizer
• Won the System Innovation Award at CIA01
  workshop, demonstrating the cinema-organizer on
  Accompli008 through GSM and iPAQ through WaveLAN.
• Next challenges
• Complete applications (Integration of generic
  services, legacy systems and databases)
• Heavy testing of the applications
• Prepare training and evaluation materials (user
  documentation, interviews)
• Run the field trials
• Evaluate the field trials

19
The future

• Agentcities.RTD and Agentcities.NET, two new
  European projects deploying a worldwide network
  of agent platforms.
• A lot of spin-off projects.
• Active research concerns
• Security, trust and privacy
• Ontology sharing
• Service clustering
• Integration with m-Commerce
• From specialization to personalization
• Social and cultural impacts.

London
Ipswich
Paris
Berlin
Dublin
Saarbruecken
Montpellier
Sendai
Lausanne
San Francisco
Lisbon
Parma
Barcelona