guest ::
TY - THES AU - Mertz, Frank TI - Efficient audio communication over heterogeneous packet networks with wireless access CY - Aachen PB - Publikationsserver der RWTH Aachen University M1 - RWTH-CONV-125664 SP - XIV, 238 S. : graph. Darst. PY - 2010 N1 - Prüfungsjahr: 2010. - Publikationsjahr: 2011 N1 - Aachen, Techn. Hochsch., Diss., 2010 AB - Communication networks are transforming towards all-IP networks with different fixed-line and wireless access technologies (e.g., DSL, UMTS, WLAN). One of the biggest challenges in such heterogeneous packet networks for the realization of real-time services (e.g., Voice over IP, music, and video streaming) is dealing with transmission impairments which include variable packet transmission delays and packet losses. Current systems utilize different techniques to combat these impairments, such as a Jitter Buffer to compensate the variance in transmission delay and reduce jitter based losses, Forward Error Correction (FEC) to recover lost frames, and Packet Loss Concealment (PLC) to estimate unrecoverable frame losses. An issue which has not yet been sufficiently addressed is how to choose the best suitable methods and parameterize them optimally for a given scenario of application and transmission channel. This dissertation develops a new methodology to analyze and determine the most suitable FEC method for a given transmission scenario. This approach aims at the best possible transmission quality by considering both signal quality and end-to-end delay. The main results include: a new channel model for packet losses which is based on the Gilbert-Elliott model and can be adapted for different packet sizes and transmission intervals; analytical determination of residual frame loss probabilities after erasure correction for different FEC methods based upon the new channel model; fair comparison of methods with different packet sizes and transmission intervals given the adaptability of the new channel model; derivation of optimal system parameters by applying the newly developed methodologies to real-life scenarios of speech and music transmission in UMTS, WLAN, and heterogeneous packet networks. Furthermore, new PLC techniques for state-of-art speech codecs are developed, which are particularly suitable for wireless transmission channels. In particular, the estimation of lost codec parameters is improved by transmission of low-rate side information in following packets. By utilizing steganographic methods, this side information can be transmitted as hidden bit stream in the encoded bits of the following frame, thereby requiring no additional data rate. KW - Internettelefonie (SWD) KW - Streaming <Kommunikationstechnik> (SWD) KW - Vorwärtsfehlerkorrektur (SWD) KW - Fehlerverdeckung (SWD) LB - PUB:(DE-HGF)11 UR - https://publications.rwth-aachen.de/record/64341 ER -