Towards a Self-healing Multi-agent Platform for Distributed Data Management. Rodríguez, A., Gómez, J., & Diaconescu, A. Towards a Self-healing Multi-agent Platform for Distributed Data Management, pages 350-354. Springer International Publishing, 2017.
Towards a Self-healing Multi-agent Platform for Distributed Data Management [pdf]Paper  Towards a Self-healing Multi-agent Platform for Distributed Data Management [link]Website  abstract   bibtex   
We demonstrate a self-healing multi-agent simulation platform for distributed data-management tasks, including data collection and synchronisation. Collective tasks can be simulated within two types of environments: uncharted terrains with various obstacles, and computing networks with different complex topologies. Agents explore their environment, collect and update local data, and exchange data with agents that they encounter, until the collective task is completed. We have previously implemented several agent exploration algorithms and evaluated their performance in terms of completion speed (essential when agents may fail) and resource overheads (essential in constrained environments). Here, we focus on the agents' ability to self-heal, via local replication, so as to ensure task completion. We focus on computing network environment, where software replication is more feasible. Envisaged applications include data management in computing clouds, distributed databases, sensor networks, robot swarms and the Internet of Things.

Downloads: 0