Stuart Chalmers

Agent-based formation of virtual organisations

Virtual organisations (VOs) are composed of a number of individuals, departments or organisations each of which has a range of capabilities and resources at their disposal. These VOs are formed so that resources may be pooled and services combined with a view to exploiting a perceived market niche. However, in the modern commercial environment it is essential to respond rapidly to changes in the market to remain competitive. Thus, there is a need for robust, agile, flexible systems to support the process of VO management. Within the CONOISE (www.conoise.org) project, agent-based models and techniques are being developed for the automated formation and maintenance of virtual organisations. In this paper we focus on the former, namely how an effective VO may be formed rapidly for a specified purpose.

CONOISE: Agent-Based Formation of Virtual Organisations

Virtual organisations (VOs) are composed of a number of individuals, departments or organisations each of which has a range of capabilities and resources at their disposal. These VOs are formed so that resources may be pooled and services combined with a view to the exploitation of a perceived market niche. However, in the modern commercial environment it is essential to respond rapidly to changes in the market to remain competitive. Thus, there is a need for robust, flexible systems to support the process of VO management. Within the CONOISE (www.conoise.org) project, agent-based models and techniques are being developed for the automated formation and maintenance of virtual organisations. In this paper we focus on a critical element of VO management: how an effective VO may be formed rapidly for a specified purpose.

Frugal Sensor Assignment

When a sensor network is deployed in the field it is typically required to support multiple simultaneous missions, which may start and finish at different times. Schemes that match sensor resources to mission demands thus become necessary. In this paper, we consider new sensor-assignment problems motivated by frugality, i.e., the conservation of resources, for both static and dynamic settings. In general, the problems we study are NP-hard even to approximate, and so we focus on heuristic algorithms that perform well in practice. In the static setting, we propose a greedy centralized solution and a more sophisticated solution that uses the Generalized Assignment Problem model and can be implemented in a distributed fashion. In the dynamic setting, we give heuristic algorithms in which available sensors propose to nearby missions as they arrive. We find that the overall performance can be significantly improved if available sensors sometimes refuse to offer utility to missions they could help based on the value of the mission, the sensors remaining energy, and (if known) the remaining target lifetime of the network. Finally, we evaluate our solutions through simulations.

Sensor-Mission Assignment in Constrained Environments

When a sensor network is deployed in the field it is typically required to support multiple simultaneous missions, which may start and finish at different times. Schemes that match sensor resources to mission demands thus become necessary. In this paper, we consider new sensor-assignment problems motivated by frugality, i.e., the conservation of resources, for both static and dynamic settings. In the most general setting, the problems we study are NP-hard even to approximate, and so we focus on heuristic algorithms that perform well in practice. In the static setting, we propose a greedy centralized solution and a more sophisticated solution that uses the Generalized Assignment Problem model and can be implemented in a distributed fashion. In what we call the dynamic setting, missions arrive over time and have different durations. For this setting, we give heuristic algorithms in which available sensors propose to nearby missions as they arrive. We find that the overall performance can be significantly improved if available sensors sometimes refuse to offer utility to missions they could help, making this decision based on the value of the mission, the sensors remaining energy, and (if known) the remaining target lifetime of the network. Finally, we evaluate our solutions through simulations.

GraniteNights – A Multi-agent Visit Scheduler Utilising Semantic Web Technology

This paper describes a multi-agent system, GraniteNights, modelled on the Agentcities project “evening agent” scenario. GraniteNights allows a user to plan an evening’s entertainment in the city of Aberdeen, Scotland. The application fuses agent and Web technology, being viewed as an agent-based Web service. In particular, Semantic Web standards are used to a great extent in delivering the service. The paper argues that, in fact, the Semantic Web standards are more important for this type of application than the agent standards. A key feature of the application is component re-use: GraniteNights attempts to reuse without modification existing ontologies wherever possible; it also is comprised of a number of generic and thus wholly-reusable agents, including a user profiling agent and a constraint-based scheduler. The system is open in the sense that most of the individual agents can be invoked directly by external agent platforms, without going through the Web interface.

Experience in Using RDF in Agent-Mediated Knowledge Architectures

We report on experience with using RDF to provide a rich content language for use with FIPA agent toolkits, and on RDFS as a metadata language. We emphasise their utility for programmers working in agent applications and their value in Agent-Oriented Software Engineering. Agent applications covered include Intelligent Information Agents, and agents forming Virtual Organisations. We believe our experience vindicates more direct use of RDF, including use of RDF triples, in programming knowledge architectures for a variety of applications.

Commitment Management through Constraint Reification

In a virtual organisation (VO), a group of cooperating agents may offer resources (or services) consisting of all, or part, of the sum of their individual contributions. Some of these resources may already be in use for certain time periods (existing commitments), and these existing commitments may vary in value to the VO. Using constraint reification and cumulative scheduling methods, we investigate ways in which agents may manage their resources and existing commitments when faced with a decision to take on new commitments. We describe a technique that allows an agent to intelligently construct satisfiable permutations consisting of existing and new commitments, and use preference and quality information to choose between these permutations. We show how constraint reification is used to model whether commitments are breakable/re-negotiable and how this influences the permutations available to the agents.

CONOISE-G: agent-based virtual organisations

The ability to create reliable, scalable virtual organisations (VOs) on demand in a dynamic, open and competitive environment is one of the challenges that underlie Grid computing. In response, in the CONOISE-G project, we are developing an infrastructure to support robust and resilient virtual organisation formation and operation. Specifically, CONOISE-G provides mechanisms to assure effective operation of agent-based VOs in the face of disruptive and potentially malicious entities in dynamic, open and competitive environments. In this paper, we describe the CONOISE-G system, outline its use in VO formation and perturbation, and review current work on dealing with unreliable information sources.

Assisting domain experts to formulate and solve constraint satisfaction problems

Constraint satisfaction is a powerful approach to solving a wide class of problems. However, as many non-experts have difficulties formulating tasks as Constraint Satisfaction Problems (CSPs), we have built a number of interfaces for particular kinds of CSPs, including crypt-arithmetic problems, map-colouring problems, and scheduling tasks, which ask highly focused questions of the user, c.f., the earlier MOLE/MORE, and SALT knowledge acquisition systems. Information from each of these interfaces is then transformed initially into a structured format which is semantic web compliant and is secondly transformed into the format required by the generic constraint satisfaction problem solver. When this problem solver is run, the user is either provided with solution(s) or feedback that the problem is underspecified (when many solutions are feasible) or over-specified (when no solution is possible). The system has 3 distinct phases, namely; information capture, transformation of the information to that used by a standard problem solver, and thirdly the solving and user feedback phase.

Supporting Virtual Organisations Using BDI Agents and Constraints

Virtual organisations underpin many important activities in distributed computing, including e-commerce and e-science. This paper describes a new technique by which software agents can intelligently form virtual organisations to meet some pre-specified requirements. Our approach builds on work in BDI agents and constraint satisfaction techniques. Using a realistic service-providing scenario, we show how an agent can use constraint solving techniques to explore possible virtual organisation alliances with other agents, based on its beliefs and desires. The agent can choose the best among several possible virtual organisations to form in order to meet a customers requirements. Our approach is to use a deliberative process to construct possible worlds, each corresponding to a potential virtual organisation, and each configured using constraint satisfaction techniques. We also show how an agent can take account of pre-existing virtual organisation relationships in its deliberations.