Designing Device-Aware Adaptive Video Streaming Algorithm

Project Title

Designing Device-Aware Adaptive Video Streaming Algorithm

Project Area of Specialization Computer ScienceProject Summary

To demonstrate the usefulness of device-awareness, we implement DAVS in the dash.js framework. Our initial results show that DAVS can improve FPS and minimize crashes. Our work extends existing bitrate algorithms to consider device bottlenecks and provides insights about how applications can be designed to run better in low-memory scenarios.

Project Objectives

The rising demand for high-resolution videos and high video frame rates are leading to increasing pressure on the memory available in mobile devices. In this work, we analyze how and when memory usage on a device impacts video performance. We propose and evaluate DAVS, a video client module that adapts the playback buffer size and the video bitrate based on device memory pressure.

Project Implementation Method

The design of DAVS is based on two key goals:

(a) to prevent player crashes (whenever possible) and (b) to improve FPS.

In DAVS, we leverage the following insights. When the device is the bottleneck (because of memory pressure), we can select a high bitrate by reducing the playback buffer size. For example, when we pick a 1080p video and an 8s playback buffer, we can reduce frame drops and reduce crashes. However, when the network is the bottleneck, we may need a larger buffer to avoid rebuffering, and a lower bitrate can be selected to optimize the video QoE.

Benefits of the Project

Internet video accounts for over 75% of global Internet traffic with mobile video contributing 62% to the world-wide video traffic share. Driven by rising demands for high-quality videos, we are witnessing two key trends: (i) support for increasingly high-resolution videos (e.g., 1080p (HD) and 2160p (4K) videos, both which are supported by YouTube and Netflix) and (ii) support for higher video frame rates (e.g., 48 fps, 60 fps). As higher quality videos consume
more memory, these trends are leading to increasing pressure on memory available in mobile devices such as smartphones. To demonstrate the usefulness of device-awareness, we implement DAVS in the dash.js framework. Our initial results show that DAVS can improve FPS and minimize crashes.

Technical Details of Final Deliverable

The final deliverable will be a report/Research paper that will demonstrate how can we improve FPS while maximizing other QoE metrics.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Education Core Technology Artificial Intelligence(AI)Other Technologies Cloud InfrastructureSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
Low-end and High End Devices	Equipment	2	35000	70000
Travel and Other Expenses	Miscellaneous	1	10000	10000