Game Theory for Networks. 6th International Conference, GameNets 2016, Kelowna, BC, Canada, May 11-12, 2016, Revised Selected Papers

Research Article

Distributed Sharing of Base Stations for Greening: A Population Game Approach

Download
202 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-47509-7_8,
        author={Soohwan Lee and Yung Yi},
        title={Distributed Sharing of Base Stations for Greening: A Population Game Approach},
        proceedings={Game Theory for Networks. 6th International Conference, GameNets 2016, Kelowna, BC, Canada, May 11-12, 2016, Revised Selected Papers},
        proceedings_a={GAMENETS},
        year={2017},
        month={1},
        keywords={Greening Base station sharing Population game theory},
        doi={10.1007/978-3-319-47509-7_8}
    }
    
  • Soohwan Lee
    Yung Yi
    Year: 2017
    Distributed Sharing of Base Stations for Greening: A Population Game Approach
    GAMENETS
    Springer
    DOI: 10.1007/978-3-319-47509-7_8
Soohwan Lee1,*, Yung Yi1,*
  • 1: KAIST
*Contact email: soohwan@kaist.ac.kr, yiyung@kaist.edu

Abstract

Towards better QoSs and larger market share in highly competitive cellular network market, many mobile network operators (MNOs) aggressively invest in their base station (BS) deployment. As a result, BSs are densely deployed and incur a lot of energy consumptions, resulting in a large portion of operation cost. To save energy consumption, sharing BSs among different MNOs is a promising approach, where each user can be served from any BSs regardless of his or her original subscription, i.e., roaming. In this paper, we address the question of how many users should be roamed in a distributed manner with the goal of some sense of optimality. To answer this question, we take a population game approach, where we model flow-level dynamics of traffic and define an user association game among different MNOs. We prove that the game is an exact potential game with We develop a distributed algorithm that converges the NE (which is a socially optimal point) that can be used as a light-weight, dynamic user association algorithm.