The Beauty of Consistency in Radio-Scheduling Decisions

Radio scheduling is a well-studied problem that has challenged researchers throughout the last decades. However, recent findings that stem from the Quality of Experience (QoE) domain come to give a new perspective to traditional radio scheduling approaches. In this paper, we take advantage of recent subjective results regarding the impact of throughput fluctuations on the QoE of interactive applications and revisit well-known scheduling algorithms. By quantifying the impact of traditional radio schedulers on user-perceived QoE, we manage to draw new conclusions regarding the radio scheduling problem, such as the importance and impact of consistency of the resource allocation decisions on the users' QoE. As main result, fair algorithms inherently seem to be more consistent than greedy ones, providing less throughput fluctuations and, thus, better QoE. Based on this outcome, we propose a new scheduling approach, which further improves end- users' QoE by moderating throughput fluctuations.