||Evaluating DASH and TCP Interactions with TEACUP Testbed
||EN615, Level 6, EN Building
||Video streaming is rapidly dominating the Internet traffic, with Netflix and YouTube accounting for almost 50% of the peak download traffic on fixed networks in North America in 2014. Dynamic Adaptive Streaming over HTTP (DASH) has recently became an ISO/IEC MPEG and 3GPP standard for live and on-demand video streaming services. TCP (Transmission Control Protocol) is often considered to be most efficient for bulk (long-lived) transfers. However, the emergence of DASH over TCP complicates things by behaving like a long string of consecutive short-flows. The interactions between DASH and TCP open up some interesting areas ripe for research. For example, optimising the TCP inner control loop that interacts with the Adaptive Bit Rate (ABR) algorithm outer loop to provide maximal Quality of Experience (QoE) without causing collateral damage to other traffic presents many challenges.
Hence, we extended the capability of TEACUP (TCP Experiment Automation Controlled Using Python) software. This extension allows us to run repeatable DASH experiment trials with different TCP algorithms over a range of emulated network path conditions, bottleneck rate limits and active queue management (AQM) schemes with industry-compliant DASH clients.
This talk presents the motivation of our work, the architecture of the TEACUP-DASH extension, an evaluation of the impact of different TCP algorithms and AQMs on the perceived QoE in an emulated home network, and possible future research directions.
||Jonathan has received his Bachelor of Engineering (Telecommunication and Network Engineering) degree with First Class Honours from Swinburne University of Technology, Melbourne in 2014. He is currently a PhD candidate at the Centre for Advanced Internet Architectures (CAIA).