Large networks
Wireless networks are the key enabling technology to dematerialization and virtual ubiquity and play a fundamental role in sustainable development. This is fueling studies of communication paradigms whose complexity is incommensurably higher than the classical single cell wireless model: ad hoc and sensor networks, cognitive radios, femto and small cells networks, massive MIMO systems, to mention only a few.
Understanding, design, and analysis of the information flow in these challenging scenarios is complicated by
- Uncertainty of the network topology;
- Uncertainty of the channel state;
- Correlation between information sources, especially in sensor networks;
- Presence of feedback channels due to the noisy and unreliable characteristics of the channel;
- Scalability and complexity issues.
The Mobile Communication team is deeply involved in the investigation of large networks from a manifold perspective:
- Global network approach;
- Cross-layer perspective;
- Network component design.
Very interestingly, interactions in these telecommunication systems are comparable to the interactions of matter particles. Then, theories borrowed from branches of physics for the investigation of macroscopic matter properties based on microscopic particles’ interactions are very relevant for the study of large networks. Surprisingly, they can be successfully applied to the analysis of large networks to model both topological and propagation uncertainties.
Indeed, perculation theory, describing the flow of a fluid through a porous material, has been utilized to analyze the flow of information through a network with randomly distributed nodes.
Quantum and statistical mechanics provided mathematical tools such as random matrix theory and replica methods to investigate communications among a large number of terminals through uncertain channels.
The averaging capabilities of these mathematical tools allow for very insightful analyses and neat and powerful descriptions of large communication systems in terms of few macroscopic parameters.