Analyzing and absorbing cross-layer header overhead of video data from end-to-end viewpoint

Chu Chuan Lee, Pao Chi Chang

Research output: Contribution to journalArticlepeer-review

Abstract

Regarding IP-based video applications over wireless networks, the multi-layer header overhead may significantly affect the estimation of target video encoding bit rate and the effective throughput of wireless network. Based on the existing header structure of video packets, this study intends to deal with the header overhead problem from the end-to-end viewpoint. This paper first proposes a simple yet robust closed-form that can determine accurately and timely the optimal video payload length at the video sender based on the current wireless channel condition. The contribution can effectively improve the WLAN throughput and enhance the error resilience effect of scalable video data simultaneously. This study further explores the impact of multi-layer header overhead to the video coding work and proposes a Dynamic Header Overhead Accommodation (DHOA) scheme, which is executed in the video compression layer, to adjust dynamically the available video encoding bits for accommodating the header overhead in advance. The contributions of this paper are robust for various IP implementations such as IPSec (IP Security) over different 802.11 standards. Analytical and simulation results verify the accuracy and effectiveness of the proposed closed-form and header accommodation method. Using DHOA, the bandwidth mismatch between the actual bandwidth demand of packetized video data and the available network bandwidth is no more than 1.1% regardless of the packet sizes used in this paper.

Original languageEnglish
Pages (from-to)4360-4367
Number of pages8
JournalIEICE Transactions on Communications
VolumeE88-B
Issue number11
DOIs
StatePublished - Nov 2005

Keywords

  • Data packetization
  • Multimedia application
  • Video streaming
  • Wireless network

Fingerprint

Dive into the research topics of 'Analyzing and absorbing cross-layer header overhead of video data from end-to-end viewpoint'. Together they form a unique fingerprint.

Cite this