Self-adapting decentralized systems for the Future Internet

The DELFIN project is focused on the design of self-adapting decentralized systems for the Future Internet. The network of the future is facing serious challenges due to an enormous increase in traffic, computing power and a miriad of novel devices connected to the network. For this reason, a new generation of distributed systems must provide extreme scalability and adaptivity to variable node and network conditions. We aim to cover the whole feedback loop of adaptive systems including: sensing and analysis, control and self-regulation, and finally system adaptation and reaction.

First of all, we will devise novel network and overlay management and monitoring sensors to obtain essential parameters like node availability, load, stability, churn, routing and topology related information.

The use of the information about the graph-theoretical structure of the logical overlay network will help to determine how to place the sensors so that a maximum coverage of the network can be achieved using the minimal number of resources. Furthermore, we will perform data analysis using cognitive techniques like probabilistic algorithms and heuristics (GRASP, genetic algorithms, etc).

We will then design novel regulation and incentive systems (self-management) aiming to optimize the distributed resource allocation problem. In this line we will use different techniques such as economic models, random and epidemic protocols, and trust and reputation systems. All these techniques will benefit from the information obtained by the sensing modules.

Finally, we will validate the self-adaptive algorithms in a variety of environments including: distributed peer-to-peer storage systems, distributed application deployment infrastructures, voluntary cloud computing networks, and large-scale collaboration systems. The validation phase will include both simulation and large-scale experimentation testbeds.

Back to projects