Improving the quality of experience with size-based and opportunistic scheduling

In cellular networks, the Shortest Remaining Processing Time first (SRPT) principle, which is known to be optimal for sharing jobs on a single-server system, can deliver significant advantages over conventional scheduling algorithms. Transmissions are finished earlier, which improves user experience and enables operators to allow more users in their networks. However, this comes at the cost of reduced cell throughput and a penalty for large objects in peak traffic situations. We therefore propose an extension to SRPT that combines it with opportunistic scheduling and allows to tradeoff the advantages of both sides. We evaluate and compare the performance of SRPT and conventional schedulers by simulation with a realistic traffic scenario. An important metric is the transmission duration of application layer objects as it is central for the users´ Quality of Experience (QoE). The results show that the proposed scheduler provides the superior QoE of SRPT for short, interactive transmissions while keeping the rate reduction for larger traffic objects at an acceptable level.