Enhanced Live Video Streaming over Integrated LTE Wi-Fi Architecture

Patidar, Tarun and Tamma, Bheemarjuna Reddy (2019) Enhanced Live Video Streaming over Integrated LTE Wi-Fi Architecture. Masters thesis, Indian institute of technology Hyderabad.

Text
Mtech_Thesis_TD1433_2019.pdf
Restricted to Repository staff only until 2 January 2020.
Download (2MB) | Request a copy

Abstract

The demand for video streaming over mobile devices has significantly increased over the years. Video is the largest constituent of mobile data traffic. Live video is one of the crucial and rising categories of video. It involves real-time communication between two users, thus has a strict low latency requirement. Current multipath techniques like Multipath Real-Time Transport Protocol (MPRTP) and Multipath Transmission Control Protocol (MPTCP) are not optimized to meet this requirement. This is especially true when they are operating over a wireless medium, whose channel conditions change very rapidly and unpredictably. These multipath techniques are unable to adapt quickly to the dynamic channel conditions. LTE WLAN Aggregation (LWA) is capable of dealing with this problem since it makes its decisions at the edge of the network. In this work, we propose the bOunded delAy baSed steerIng with timelineSs (OASIS) framework over LWA for streaming live video with low latency. The OASIS framework comprises of a novel traffic steering algorithm, BOunded deLay based sTeering (BOLT). BOLT calculates the maximum traffic load which can be transmitted on each of the links (LTE and Wi-Fi) before the packets experience higher delay i.e., above a certain delay threshold. OASIS also comprises of a timeliness model, which prioritizes the packets belonging to Intra-coded picture (I-frames) over other packets. OASIS is implemented and tested in NS-3. We use NS-3 in emulation mode which allows us to stream real videos over the simulated topology. OASIS is compared with MPRTP, LWA, LTE, and Wi-Fi. We show that the proposed framework performs 1.4× better than MPRTP. Also, it outperforms all the other comparative techniques when tested under different scenarios. These scenarios are designed to test the different techniques under rapidly changing wireless medium. We also show that the quality of the video improves with the inclusion of the timeliness model. The merits of proposed BOLT algorithm is tested on both UDP and TCP. We use Dynamic Adaptive Streaming over HTTP (DASH) to stream video on BOLT over LWA and show that it outperforms regular LWA, Wi-Fi only, and LTE only.

[error in script]

IITH Creators: