A learning-based approach to energy efficiency maximization in wireless networks

This work develops a learning-based framework for energy-efficient power control in multi-carrier wireless networks. The problem is formulated as the maximization of the network global energy efficiency, defined as the ratio between the network sum-rate and the total consumed power, and is tackled by a novel approach which merges tools from learning, non-cooperative game theory, and fractional programming theory. The proposed algorithm is provably convergent, enjoys near-optimal performance, while requiring a much lower complexity than previous alternatives.