Madkit and Ekserva

1
Madkit and Ekserva

• Jinqiao Yu
• Siplab
• CERC
• The Lane Department of CSEE
• West Virginia University
• Morgantown, WV 26505

2
What will I do?

• A brief introduction about Madkit
• Play with it
• Discussion with you about it and how it can fit
  in Ekserva

3
What is Madkit?

• Multi-agent platform for developing and running
  application based on an organizational oriented
  paradigm
• Uses agents, groups and roles as the basic
  standpoint for building complex applications
• Does not enforce any consideration about the
  internal structure of agents
• A distributed platform which allows for the
  development of efficient distributed applications
  
• MadKit is built around the concept of
  micro-kernel and agentification of services.

4

• MadKitis a free software which comes with a mix
  of GPL/LPGL licenses
• 100 pure Java

5
Install Madkit

• Required software
• Java 1.2 JVM or above
• Windows (98/2000/XP/NT), Linux (x86/PC) and MacOS
  X
• JessAgent needs Jess6.0
• Madkit.org

6

• Starting Madkit
• Desktop standard agent launching and development
  environment
• Madkit applet kernel as an applet embedded into
  HTML
• Console booter
• Madkit launcher

7
Madkit Desktop Environment

• Starting the desktop
• Double click icon (Windows)
• Command ./runmadkit (Unix)
• Run scripts

8
Desktop Quickview
9
Desktop Quickview

• Directory explorer
• Default autoload
• Click leaves will launch applications
• Toolbar
• Agent pane
• Output pane

10
Launching and observing agents

• Launching
• Click one of the buttons of the agent toolbars
• Double click leaves representing agents
• Click on one of the item in the Agents Menu
• jEdit to launch scripted agents
• Use language editor

11

• Observing
• GroupObserver
• PropertyBox

12
Configure the desktop

• Configuration file- desktop.ini
• Three ways to add new classes of agents in a
  desktop
• Add jar archive to the autoload directory
• Add a button in the deskop
• Open the configuration file

13
Using Madkit in a distributed Environment

• Each MadKit kernel may act as a node of a complex
  distributed environment
• Connections betweens kernels are done by a
  specific agent called the Communicator
• When multiple kernels are connected, the
  resulting MadKit network forms a distributed
  platform
• Agents may be used to connect users, share MadKit
  applications, services, and documents of any kind

14
Community

• Community
• A group of connected Madkit kernel
• Used to partition a whole network to
• Speed it up
• Propose a conceptual repartition of resources and
  services
• Provide a framework for securizing distributed
  applications

15
Communities with different kernels
16
Madkit host container

• Desktop container
• Console
• Console.bat or console.sh
• Applet

17
Madkit and Ekserva

• My particular concerns about Madkit
• Madkit Peer to peer
• Each kernel is directly connected to all other
  kernels
• Not possible to build large networks with this
  schema
• Madkit claims a new version of communicator will
  be availabe soon but when?

18
Use of Madkit

• For instance
• Distributed Classroom Application can be a
  community
• Each Specific Classroom can be a group
• Instructor can be the manager of that group

19

• What are the changes needed to be made?