Ph.D Scholarship Programme "Heraclitus II"

In the last decade, the energy needs of all sectors of our modern societies have risen significantly. Thus, energy efficiency has become a major target of the research community. The ongoing research efforts are focused into two main threads, i) improving energy efficiency of individual devices and ii) optimizing the efficiency and reliability of the power grid.

In particular, total energy dissipation by communication devices has been shown to amount to a significant portion of a nation’s power profile, motivating efforts of per device energy economy. In this direction, this thesis investigates the problem of medium access for wireless devices that support sleep modes and provides a characterization the energy–throughput tradeoff for contention based systems. Towards the goal of energy efficiency, the problem of task scheduling in a cloud computing environment is also considered. Indicatively, cloud datacenters in USA currently account for 3% of total annual electricity consumption and is growing rapidly. We propose VM migration mechanisms that lead to significant energy savings by enabling us to turn off unused servers, without sacrificing the quality of experience of the users.

Towards a more efficient power grid, we develop dynamic pricing mechanisms that motivate users to shift or curtail their demands. In particular, we introduce a hierarchical market model for the smart grid where a set of competing aggregators act as intermediaries between the utility operator and the home users. The increased negotiation power of the aggregators is exploited so as to maximize the financial benefits of the home users.

The common ground of these problems is the autonomous nature of the participating entities. Each user acts independently towards maximizing her own utility, a strategy that generally does not coincide with the socially optimal. Thus, we use optimization and game theory techniques to model and analyse the performance of such systems.