Cost optimized multipath scheduling in 5G for Video-on-Demand traffic

This paper evaluates the limitations of existing scheduling algorithms when video-on-demand traffic is transported in multipath scenarios, and proposes a new scheduling algorithm called cost-optimized multipath (COM). The new algorithm is designed to decrease the mobile network operators' cost of the delivery of bursty video-on-demand traffic over multipath networks access. Local and Internet connected testbeds, as well as trials with real cellular customers have been deployed to analyse the video performance over MPTCP-based multipath. The results clearly demonstrate the impact the bursty nature of video-on-demand traffic has on the scheduling decisions in multipath scenarios, when traditional latency-based or cheapest-path-first schedulers are deployed. Based on the testbed and trial results, this paper presents the design of a new simple and scalable scheduling algorithm. The paper describes the typical use cases and shows preliminary testbed results, clearly demonstrating the cost benefits of the new algorithm, and indicating that the right balance between the user QoE and the operator cost can be achieved for the video traffic.