A Markovian Design of Bi-Directional Robust Header Compression for Efficient Packet Delivery in Wireless Networks

Wenhao Wu*, Zhi Ding

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)

Abstract

As a major tool for improving transport efficiency by reducing header redundancy, robust header compression (ROHC) plays an important role in modern packet-switched wireless networks. However, widespread ROHC deployment is in sharp contrast to extremely limited number of research works on ROHC analysis and design optimization. In this paper, we investigate a novel trans-layer approach in designing a bi-directional ROHC under unreliable wireless channel conditions. We propose a novel ROHC compressor design based on a new formulation in terms of a partially observable Markov decision process (POMDP). This new formulation robustly explores the lower protocol layer signaling to optimize the compressor actions dynamically on the header level and feedback polling. Our design approach improves the transmission efficiency and curtails the required ROHC feedback overhead. Furthermore, to reduce the complexity of our optimized POMDP design, we propose a low-complexity suboptimal ROHC compressor. Our novel trans-layer designs achieve more flexible trade-offs between transmission efficiency and feedback overhead than existing ROHC compressors. They also demonstrate a substantial performance improvement under poor channel conditions and long feedback delay.

Original languageEnglish
Article number8497033
Pages (from-to)20-33
Number of pages14
JournalIEEE Transactions on Wireless Communications
Volume18
Issue number1
DOIs
Publication statusPublished - Jan 2019
Externally publishedYes

Keywords

  • Packet networks
  • ROHC
  • compression
  • decision process
  • packet header
  • partial observation

Fingerprint

Dive into the research topics of 'A Markovian Design of Bi-Directional Robust Header Compression for Efficient Packet Delivery in Wireless Networks'. Together they form a unique fingerprint.

Cite this