- Agent-based Simulation (1) (remove)
- Towards the Governance of Open Distributed Grids - A Case Study in Wireless Mobile Grids (2011)
- New networking technologies such as wireless mobile grids and peer-to-peer middleware are examples of a growing class of open distributed systems whose strength is the absence of a central controlling instance and which function through the cooperation of autonomous entities that voluntarily commit resources to a common pool. The social dilemma in such systems is that it is advantageous for rational users to access the common pool resources without making any commitment of their own. This is commonly known as “free-riding”. However, if a substantial number of users followed this selfish strategy, the system itself would fail, depriving all users of its benefits. In this dissertation, we demonstrate how governance decisions can induce cooperation in such systems and how normative frameworks in combination with multi-agent system simulations can be successfully employed to analyse their effects, even at an early development stage. We show that our approach is not only practical and powerful, but also easily accessible. We demonstrate its functionality by implementing a prototype to explore the impact of enforcement mechanisms on wireless mobile grids, a concept which has been proposed to address the energy issues arising in the next generation of mobile phones and the networks that connect them. We also infer lessons from this example for open distributed systems in general. Simulation experiments quantify the benefits of enforcement mechanisms for wireless mobile grids. We analyse these results with respect to the costs of enforcement as well as further criteria that reflect the interests of the multiple stakeholders in the system. We conclude with some observations on how the lessons learned from both process and outcomes may be applicable to the broader context of open distributed systems. In particular, we highlight (i) the use of simulation using intelligent agents and a normative framework as a means for in silico exploration of complex systems for both business and technological objectives, and (ii) the insight offered into a range of enforcement mechanisms and a better understanding of the conditions and constraints under which they are applicable.