Emulation based BS association in wireless cellular networks

Abstract

In order to improve the efficiency of network resource utilization and reduce regional congestion, proper association of mobile stations (MSs) with serving base stations(BSs) is of crucial importance in wireless cellular networks. There have been many efforts in literature on BS association problems, where most are classified closed-form approaches. In these approaches, each MS determine its association with a BS which provides maximum throughput to it, and the throughput from a BS is simply estimated by dividing average channel gain to the number of MSs in the BS. However, the closed-form approaches have limited application in the situation where a closed-form cannot be easily computed. Examples include (i) objectives of general alpha-fairness, (ii) dynamic patterns of mobility and flow arrivals/departures whose statistical nature is generally unknown. In this paper, we propose a novel emulation-based approach, called ViSE (Virtual Scheduling Emulation). ViSE emulates a virtual scheduling that tries to track the behavior of optimal scheduling with practical complexity and changes BS associations whenever the current association differs much from that by virtual scheduler. Due to emulation, ViSE can easily track the cases where a closed-form approach cannot easily applied. We demonstrate that in various topologies ViSE takes about 0.01% - 35% of performance gap from the optimal BS association and achieves 0.01% - 270% performance improvements with practical complexity and signaling overhead.