Tame: time window scheduling of wireless access points for maximum energy efficiency and high throughput

Abstract

Wi-Fi interface is one of the predominant energy consumers in Wi-Fi stations. Despite many researches on Wi-Fi energy management, energy wastage of Wi-Fi stations resulting from network contention among mul-tiple access points (APs) has not been widely investigated. The network contention usually occurs in two situa-tions: among an AP’s associated stations and among stations associated with multiple APs. In this paper, we focus on the latter situation. We analyze the network contentions occur among multiple APs, and show that Wi-Fi power save mode performance could be severely affected by network contentions. In order to overcome the network contention problem, we propose a scheduling policy, Tame, to assign multiple access points into different subclusters, in each of which none of the access points have network contentions and data can be transmitted simultaneously without collision. Access points assigned into different subclusters cannot transmit data at the same time, otherwise those APs will have network contentions and energy of Wi-Fi stations can be wasted. To avoid the data transmission time overlapping, we need to turn off the MORE_DATA flag embedded in the data packets so that the APs can make the Wi-Fi stations go to sleep. Every subcluster is assigned to the same time slot to transmit data to guarantee the network throughput fairness. To further enlarge the throughput, we make use of maximum independent set algorithm and set permutation algorithm to enlarge the number of access points assigned to each subcluster and reschedule the beacon time, thus the overall system’s average throughput is enhanced. Since only during the assigned transmission time will the associated Wi-Fi stations get data and in the other time the stations keep sleeping, the Wi-Fi stations’ energy consumption is reduced dramatically. At the same time, we study the related work Sleepwell and propose Tame strategy. Compared with Sleepwell, Tame improves the throughput. We...