Mobile Agents - Computer and Information Science

4 downloads 9 Views 400KB Size Report
Jun 29, 2001 - It is suitable to describe current trends in computer ... delegated) vs. passive tools. • The term agent in computer science refers to software agent ...

Mobile Agents Haiping Xu Computer Science Department The University of Illinois at Chicago

06/29/01

Computer Science Dept., UIC

1

Introduction • The term agent comes from greek “agein”, which means to drive or to lead. • Today the term agent denotes something that producing an effect, e.g., drying agent, a shipping agent. • It is suitable to describe current trends in computer science: active instruments (to which work can be delegated) vs. passive tools. • The term agent in computer science refers to software agent.

06/29/01

Computer Science Dept., UIC

2

1

Space of Software Agents Agency

Service interactivity Application interactivity Data interactivity Representation of users Asynchrony

Message passing Remote procedure call Remote execution Weak migration Strong migration Mobility

06/29/01

Preferences Reasoning Intelligence Planning Learning

Computer Science Dept., UIC

3

Current Researches on Agents • Do not exploit all the capabilities classified by these three dimensions. • Multi-agent systems (MAS) – Execute a given task. – Use distributed but static agents. – Collaborate and cooperate in an intelligent manner.

• Mobile agents (MA) – Model agent mobility and agent coordination. – Assume very limited or even no intelligence. 06/29/01

Computer Science Dept., UIC

4

2

Evolution of the Mobile Agent Paradigm parameter (data)

Client

Server

RPC

Server

Remote Evaluation

Server-2

Agent Migration

results (data)

procedure (code)

Client results (data)

2. agent migration

Server-1 agent (code + data + state)

1. agent dispatch

Client

06/29/01

3. agent migration

4. agent migration

Server-3

Computer Science Dept., UIC

5

Degrees of Mobility transport of code + data

migration of code + data

constants parameter

migration of code + data + state

manual state encoding

mobility Code Mobility

Remote Execution Code on Demand Weak Migration

06/29/01

Strong Migration

Computer Science Dept., UIC

Agent Mobility

6

3

Why Mobile Agents ? • Software-Distribution on Demand – Easy to transport code and install packages automatically. – Code mobility simplify the management of an existing structure.

• Reduction of Communication Costs – The number of interactions. – The amount of data communicated over the network. 06/29/01

Computer Science Dept., UIC

7

Why Mobile Agents ? (continue) • Asynchronous Tasks – Asynchronous processing of requests. – Mobile device can be disconnected and reconnected.

• Scalability Due to Dynamic Deployment – A hierarchy of mobile agents can be set up. – The structure of agent hierarchy can change dynamically. 06/29/01

Computer Science Dept., UIC

8

4

A Mobile Agent System Host-A

Host-B

AVM: Φ

AVM: Θ (4)

FA: ρ

MA: β

(3)

MA: α

(1)

FA: λ

MA: α

… (2)

MA: γ



network

(1) move-request, (2) grant, (3) notify, (4) move

Agent world architecture and an example of agent migration

06/29/01

Computer Science Dept., UIC

9

Examples of Mobile Agent Systems • Mole (University of Stuttgart) – First Java-based of a mobile agent system. – Use Java as the agent programming as well as the implementation language.

• Aglets (IBM) – An aglet is a mobile java object and corresponds to mobile agents. – Support synchronous and asynchronous message passing. – Agent mobility is implemented by weak migration. 06/29/01

Computer Science Dept., UIC

10

5

Examples of Mobile Agent Systems (continue) • Agent Tcl (Dartmouth College) – An extension of the Tool Command Language (Tcl), which is a scripting language on UNIX. – Implements strong mobility.

• MARS (Univ. of Modena and Reggio Emilia) – An architecture for mobile agent coordination. – Based on a reactive tuple space model. – The tuple space is programmable.

06/29/01

Computer Science Dept., UIC

11

The MARS Architecture

06/29/01

Computer Science Dept., UIC

12

6

Academic Research Work • Model agent mobility (physical vs. logical) – Distributed join-calculus: an extension of π-calculus that introduce the explicit notions of named localities and distributed failure. – Mobile UNITY: a programming notation that captures the notion of mobility and transient interaction among mobile nodes. – MobiS: an extended version of PoliS, which is a specification language based multiple tuple spaces. – LIME: a middleware based on tuple spaces. 06/29/01

Computer Science Dept., UIC

13

Academic Research Work (continue) • Model agent communication – Knowledge Query Manipulation Language (KQML) – Foundation for Intelligent Physical Agents (FIPA) – Mobile Agent System Interoperability Facility (MASIF)

• Model agent coordination – Inter-agent coordination vs. agent-environment coordination. – Example: context-dependent coordination in MARS.

06/29/01

Computer Science Dept., UIC

14

7

Challenges • Security – Four areas: (1) inter-agent security (2) agenthost security (3) inter-host security (4) hosts and unauthorized third party. – Agent-host security: • Protect hosts from malicious (visiting) agents. • Protect agents from malicious hosts.

06/29/01

Computer Science Dept., UIC

15

Challenges (continue)

06/29/01

Computer Science Dept., UIC

16

8

Challenges (continue) • Control structures – Primitives: create, clone, and terminate agents. – To terminate agents is more complicated in the context of a hierarchy of agents.

• Transactional support – To guarantee that the agent is performed exactly once, independent of communication and node failures. – Upon failure, agent states must be made recoverable.

06/29/01

Computer Science Dept., UIC

17

My Current Research Work • Proposed a general model for intelligent software agents. – Explicitly model asynchronous message passing. – Introduce inheritance mechanism into agent-oriented software design. – Exercise behavioral analysis and verification.

• Model intelligent mobile agents (IMA). – Introduce mobility into agent-oriented software model. – Provide a framework for intelligent mobile agent.

06/29/01

Computer Science Dept., UIC

18

9

References •

N. R. Jennings, K. Sycara and M. Wooldridge, “A Roadmap of Agent Research and Development,” International Journal of Autonomous Agents and Multi-Agent Systems, 1(1): 7-38, 1998.



K. Rothermel and M. Schwehm, “Mobile Agents,” In: A. Kent and J. G. Williams (Eds.): Encyclopedia for Computer Science and Technology, Volume 40 Supplement 25, New York: M. Dekker Inc., 1999, pp. 155-176.



N. M. Karnik and A. R. Tripathi, “Design Issues in Mobile-Agent Programming Systems,” IEEE Concurrency, July-September 1998.



Tim Finin, Yannis Labrou and Yun Peng, “Mobile Agents can Benefit from Standards Efforts in Inter-agent Communication,” IEEE Communications Magazine, Vol. 36, No. 7, pp. 50-56, July 1998.



G. Cabri, L. Leonardi, F. Zambonelli, “Engineering Mobile-Agent Applications via Context-dependent Coordination,” In Proceedings of the 23rd International Conference on Software Engineering (ICSE 2001), Toronto, Canada, 2001, pp.371-380.

06/29/01

Computer Science Dept., UIC

19

Website for Mobile Agent Systems • Mole (University of Stuttgart) – http://mole.informatik.uni-stuttgart.de/

• Aglets (IBM) – http://www.trl.ibm.com/aglets/

• Agent Tcl (Dartmouth College) – http://agent.cs.dartmouth.edu/

• MARS (Univ. of Modena and Reggio Emilia) – http://sirio.dsi.unimo.it/MOON/MARS/index.html

06/29/01

Computer Science Dept., UIC

20

10

Suggest Documents