Estimation and use of load threshold for energy saving in wireless networks

Energy saving is an increasing trend in wireless communications development activities. This paper investigates how historic traffic load data can be used to estimate and set the load threshold used for switching on or off base stations or their features on a particular weekday. The use is enhanced by an abnormal day detection feature that is used to prevent undesired power on/off actions on days were users do not follow their usual behavior. Power on and off are also prohibited due to impulsive, i.e., single, observations that exceed state change thresholds. The simulation results show that the method estimated power on/off moments according to desired load level very well, especially if the load variation is limited, while preventing undesired power on/off actions.