On Efficient Packet-Switched Wireless Networking: A Markovian Approach to Trans-Layer Design and Optimization of ROHC

摘要

In packet-switched radio links, the little known Robust Header Compression (ROHC) has become an integral part of many wireless and particularly cellular communication networks. To strengthen existing schemes, this paper aims to improve ROHC performance in terms of payload efficiency for U-mode compression under poor wireless channel conditions. We first consider the parameter optimization of current ROHC systems, for which we propose a Markov compressor model suitable for realistic unidirectional (U-mode) ROHC. We present both the steady-state analysis and the transient behavior analysis of the ROHC. More generally, we propose a novel trans-layer ROHC design concept by exploiting lower cellular network layer status information to adaptively control header compression without dedicated feedbacks. Considering practical delay and inaccuracy when acquiring lower layer information, we develop a ROHC control framework in terms of a partially observable Markov decision process. Our results demonstrate the strength of our Markov ROHC compressor model in characterizing both stationary and transient behaviors, and the significant advantage of the proposed trans-layer ROHC design approach.