On a Prediction Algorithm of Arrivals of Mobile Terminal for a Green Wireless LAN System

摘要

We can reduce unnecessary electric power that unused Access Points (APs) consume in a wireless LAN system by turning off these unused APs. This turning-off operation, however, may incur regressions of user experiences, i.e., some mobile terminals may be unable to associate with turned-off APs. It is thus important to predict arrivals of mobile terminals to minimize the regressions. This paper discusses a practical algorithm to predict these arrivals using a multi-scale modeling of Poisson process. Our algorithm aims to more accurately predict arrivals of mobile terminals using a multi-scale time unit. We have then evaluated our algorithm using actual wireless LAN traces.