MESH

MESH is a research project at the University of Alcalá on the optimisation of distributed systems: communication networks, urban traffic, the Internet of Things or cyber-resilience mechanisms. Optimising them is hard for three reasons that compound each other: their scale, the hidden interrelationships among their elements and the feedback loops those relationships produce.

Classical metaheuristics scale well but are blind to the structure of the problem. Message-passing and negotiation techniques exploit that structure, but scale poorly. Graph neural networks have shown that using the structure inside the algorithm itself changes performance. MESH brings that idea to optimisation: it models each system as a multilayer factor graph and designs algorithms that natively exploit the relationships the graph expresses.

On this site we publish the project's results as they consolidate, its publications and the datasets, software and models we release so that experiments can be reproduced.