User association in cellular networks

Abstract

With increasing demands for mobile data traffic and efforts for a better QoS, many base stations (BSs) consume a lot of electrical power with huge electricity bill. Many practical solutions including determining the coverage of BS is considered to enhance the spectral efficiency. In this dissertation, we study a distributed user association policy for two-type of system, where the one is non-BS sharing, in which each mobile network operator (MNO) independently operates his own BSs, and the other one is BS sharing, in which an MNO’s BS is allowed to serve traffic from nearby users subscribing to other MNOs with being paid a certain roaming fee. For the non-BS sharing system we propose a novel user association algorithm which is well compatible with a variety of user scheduler while providing robustness to network dynamics in distributed manner. We demonstrate through extensive simulations that our proposed algorithm outperforms the existing algorithms. For BS sharing system, we investigate pricing and user association policies that assure actual gains to each MNO with saving BS power consumption by formulating a game that jointly involves strategic decision of roaming pricing and user association. We consider a time scale separation where user association decision made at a faster time scale than pricing decision and separate the game to user association game and pricing decision game. For the user association game, we prove that (i) the game is an exact potential game, (ii) there exists a unique pure Nash equilibrium, and (iii) a distributed algorithm converges to the unique Nash equilibrium. We also analyze pricing decision game and prove that (i) there exists a unique pure Nash equilibrium, and (ii) the best response dynamic converges to the Nash equilibrium. Finally, we demonstrate that there exists a significant degree of economic gain including energy saving through extensive simulations.