E-Business Models

Front Page -- Deliverable NÐ'o: 6 1

Deliverable Summary Sheet 2

1 Executive Summary 4

1.1 Purpose of the project 4

1.2 Approach, assumptions and goals 4

1.3 Implementation language and support 4

2 Introduction 4

2.1 Agents - what are they? 5

2.2 The promise 5

2.3 The reality and basic infrastructure 5

2.4 Document contents 5

2.5 Cross-reference: sections versus deliverables 6

3 Introduction to Agent Modeling and Simulation 6

4 Two examples 7

4.1 Sugarscape 7

4.1.1 Agents of the SSC system 9

4.2 Emergency Transport 10

4.2.1 Agent Communication Language 11

4.2.2 Agent Interaction Protocols 12

4.2.3 Simulation Setup - subscriptions of data 13

4.2.4 Simulation acts - negotiation for resources 14

4.2.5 Need to plan and focus on certain goals -- motivation 15

4.3 Differences - SSC versus EmT 17

5 Mozart 17

6 Design of a Multi Agent Platform 18

6.1 Base Architecture 18

6.2 Agent Base Environment 19

6.3 Extending a Base Agent 21

7 Modeling Simple Agents 22

7.1 Roles in SSC in respect to Basic Architecture 22

7.2 Use of the ABE in SSC 23

7.3 Summary 23

8 First steps towards distribution 24

8.1 Federations of agents 25

8.2 Federation setup 27

9 Support for Simulations 27

10 Programmable Behavior 28

11 Concrete Real World Simulation 28

12 Parallelisation and Distribution 28

13 Using the Simulation Environment 28

14 Concrete Studies 28

14.1 Web Word of Mouth - a detailed example 29

15 Appendix 29

15.1 Agent Base Environment 29

16 References 29

1 Executive Summary

The purpose of this document is to provide a self-contained description of the work related to the iCities project work-package number 3 (WP3), "Multi Agent System Design, Implementation and Experimentation". It is supposed to be a direct correspondence between the deliverables (reports) and the sections in this document. A cross-reference between sections and deliverables (some of them do no exist yet, but are still supplied as empty sections) is given in the introduction section.

1.1 Purpose of the project

The purpose of the iCities project is to add to the "understanding of the evolution of the Web" and to "add to the understanding of agent based computational modeling". In WP3 we focus on the latter part, based on the requirements of the web scenarios (models) being developed in other work-packages.

1.2 Approach, assumptions and goals

The task is to develop and to perform experiments with a platform or an environment for agent based simulation - in particular simulation of city-like formations on the web. The agents are assumed to be the conceptual entities, within the models, that perform some sort of actions, typically info-habitants of the Web. So what we want to study is the emergent behavior in a (simulated) web caused by the individual actions of the agents.

1.3 Implementation language and support

The simulation platform is being implemented on top of the Mozart [7] system, which features a multi-paradigm language with support for distribution and data-synchronization. One of the goals of the project is to support large scale simulations in the scale of 10K agents. To cope with such large numbers it is expected that simulations will be run on multiprocessor machines or to be distributed over a cluster of machines.

2 Introduction

The agent field has become increasingly popular in areas of research, development and applications. Because of the wide variety of directions pursued in these areas it is important do define what we mean by what we refer to as an agent. In fact the “agent-field” is not just one field, but several fields are more or less related. Fields that come to mind are: (1) Information Collection and Information Management, (2) Electronic Commerce, (3) Problem Solving, (4) Simulation, (5) Knowledge Sharing, (6) Internet, (7) Mobility and Processes, (8) Demons and Objects - just to name a few. To categorize these apples, oranges and the occasional carburetor being thrown in, is a topic in itself. (1-4) are application fields, (5) a sub-field that may be studied in its own right, (6) enabling technology, (7) a property and (8) being implementation concepts, abstractions and techniques. Add to this the different aspects of realization and the picture is getting somewhat complex.

2.1 Agents - what are they?

So what meaning do we put into the notion of an agent and how does that notion relate to the iCities project. An agent is a metaphor [10], for a conceptual entity that may be represented and implemented as some simple data structure with access functions or as an object maybe with its own thread. The entity may have some of the following properties; autonomous, reactive or proactive - sometimes refereed to as (using Distributed AI vocabulary) a 'weak' agent. A 'strong' agent is often ascribed mental abilities such as beliefs, attitudes and means for re- evaluating and re-focusing their goals. This means that a powerful programming language and the programming constructs of