Date

2018

Major

Electrical Engineering (B.S.)

Document Type and Release Option

Thesis (open access)

Faculty Mentor

Rami J. Haddad

Abstract

In this work, we propose the use of Coarse Grain Scalable (CGS) and Medium Grain Scalable (MGS) H.264/AVC video to optimize video playback on passive optical networks (PONs) by investigating network performance metrics such as data delay, video delay, and video delay jitter. Video playback is improved by sequentially dropping layers of scalable video. Dropping just a single CGS enhancement layer results in improvements of up to 57% for both data and video delay. However, video delay jitter benefits the most with an improvement ranging from 47% to 87%. Surprisingly, dropping subsequent CGS enhancement layers does not significantly improve the PONs performance. In order to remedy this effect, our focus switched to employing the H.264/AVC MGS video standard. Though video traffic delay is the primary object of optimization in this work, the proposed algorithm’s impacts on other network performance metrics such as data traffic delay and video traffic delay variance (jitter) are analyzed as well. Video playback is improved by employing an adaptive scalable video layer dropping algorithm which drops a progressively larger number of scalable video layers as network utilization increases as measured by the moving average of the video packet delay. The influence of the algorithm's three parameters on its performance is investigated in detail, and the results of the optimized adaptive dropping algorithm are compared to baseline static dropping algorithm.

Share

COinS