Affinity propagation for energy-efficient BS operations in green cellular networks

This paper develops a distributed strategy to identify an energy-efficient base station (BS) network configuration for green cellular networks. During off-peak periods where traffic demands are only a fraction of the peak-time traffic demands, a subset of BSs is switched off to minimize operational energy consumption without affecting service to any of network users. To this end, we formulate a combinatorial optimization of jointly determining BS switching and user association. This formulation, however, requires a computationally demanding task as the population of the network grows. To resolve these challenges, we introduce a graphical-model approach to the optimization formulation and derive a distributed algorithm based on affinity propagation, which is a message-passing algorithm developed for data clustering in data-mining techniques. The proposed algorithm operates via simple local information exchanges among users and BSs and provides a very efficient solution for energy-saving management with low computational costs. We also present a green protocol that transforms commercial cellular networks into green radio networks using the proposed algorithm. Simulation results verify that the developed solution significantly improves the energy savings and resource utilization in the network.