Presentation Title

Scalable-Video-Based Dynamic Bandwidth Allocation Algorithms for EPONs

Location

Nessmith-Lane Atrium

Session Format

Poster Presentation

Research Area Topic:

Engineering and Material Sciences - Electrical

Abstract

With the global IP traffic increasing exponentially every year to approach 2 zettabytes (10^21) by the year 2019, Cisco predicted that 80% of this total traffic will be IP video. North American consumers alone are projected to use almost 24 exabytes of internet video traffic. This growth in IP traffic is mainly due the increase of IP devices and the advances in our infrastructure optical IP networks. One of the most recent improvements in our IP networks are the all-optical access networks that connects the users to the service provider. The legacy copper access networks such as xDSL and Cable are the main bottleneck in all IP networks. Passive optical networks (PONs) are the new standard all-optical access networks that are being implemented worldwide to eliminate the bottleneck. Google-Fiber and Verizon-FiOS are examples of commercial PONs. In this research, an algorithm is developed to scale the resolution of video streams in PONs to reduce congestion while ensuring global fairness among all users. This algorithm needs to be computationally streamlined since PONs are by nature very cost-sensitive. It utilizes the centralized nature of the medium access control (MAC) protocol of PONs to achieve such cooperative scaling of video streams. PON simulation software aids the cost-effective development of the algorithm. Video traces created using the H.264/AVC video codec are used to ensure the algorithm’s real-world applicability. This procedure is readily extendable to other video codecs such as MPEG-4 Part 2, H.265/HEVC, and VP9. This cooperation managed by the optical line terminal at the service provider end will help ensure fairness in scaling users’ video and their dynamic bandwidth allocation. The fixed and exhaustive using queue size prediction grant sizing techniques respectively afford PONs perfect fairness and high utilization. Both characteristics are highly desirable in access networks, so this algorithm draws a fine line between the two ideals and applies a highly-effective compromise.

Presentation Type and Release Option

Presentation (Open Access)

Start Date

4-16-2016 2:45 PM

End Date

4-16-2016 4:00 PM

This document is currently not available here.

Share

COinS
 
Apr 16th, 2:45 PM Apr 16th, 4:00 PM

Scalable-Video-Based Dynamic Bandwidth Allocation Algorithms for EPONs

Nessmith-Lane Atrium

With the global IP traffic increasing exponentially every year to approach 2 zettabytes (10^21) by the year 2019, Cisco predicted that 80% of this total traffic will be IP video. North American consumers alone are projected to use almost 24 exabytes of internet video traffic. This growth in IP traffic is mainly due the increase of IP devices and the advances in our infrastructure optical IP networks. One of the most recent improvements in our IP networks are the all-optical access networks that connects the users to the service provider. The legacy copper access networks such as xDSL and Cable are the main bottleneck in all IP networks. Passive optical networks (PONs) are the new standard all-optical access networks that are being implemented worldwide to eliminate the bottleneck. Google-Fiber and Verizon-FiOS are examples of commercial PONs. In this research, an algorithm is developed to scale the resolution of video streams in PONs to reduce congestion while ensuring global fairness among all users. This algorithm needs to be computationally streamlined since PONs are by nature very cost-sensitive. It utilizes the centralized nature of the medium access control (MAC) protocol of PONs to achieve such cooperative scaling of video streams. PON simulation software aids the cost-effective development of the algorithm. Video traces created using the H.264/AVC video codec are used to ensure the algorithm’s real-world applicability. This procedure is readily extendable to other video codecs such as MPEG-4 Part 2, H.265/HEVC, and VP9. This cooperation managed by the optical line terminal at the service provider end will help ensure fairness in scaling users’ video and their dynamic bandwidth allocation. The fixed and exhaustive using queue size prediction grant sizing techniques respectively afford PONs perfect fairness and high utilization. Both characteristics are highly desirable in access networks, so this algorithm draws a fine line between the two ideals and applies a highly-effective compromise.