Link Error Analysis and Modeling for Cross-Layer Design in UMTS Mobile Communication

Particularly in wireless mobile communications, link errors severely affect the quality of the services due to the high error probability and the specific error characteristics (burst errors) in the radio access part of the network. In this thesis it is shown that a thorough analysis and the appropriate modeling of the radiolink error behaviour is essential not only to evaluate and optimize the higher layer protocols and services. It is also the basis for finding network-aware cross-layer processing algorithms which are capable of exploiting the specific properties of the link error statistics (e.g. the predictability). This thesis presents the analysis of the radio link errors based on measurements in live UMTS (Universal Mobile Telecommunication System) radio access networks. It is shown that due to the link error characteristics basically two scenarios have to be distinguished: static and dynamic (regardless of which kind of mobility). Furthermore, the analysis shows that apart from fading effects, also the quality-based power control mechanism induces correlation to the link error characteristics, resulting in error bursts, error clusters and the predictability of the link errors. It is presented that ‘well-known’ classical error models are not capable of describing the measured specific error characteristics of the UMTS DCH (Dedicated Channel) properly. Especially the error predictability cannot be represented at all due to the geometric distributed output of the models. Thus, a new modeling approach based on semi-Markov models as well as (Markov modulated) Weibull renewal processes is presented. Simple estimators for future link error probabilities are derived from these models. Based on measurements, the dynamic resource allocation of the UMTS is analysed and models for the dynamic bearer type switching are proposed which together with the link error models can provide a proper representation of the dynamic radio link behaviour. In this thesis, novel cross-layer algorithms are presented which make use of the discovered predictability of the link errors and consider the service characteristics. It is shown that high gain in the quality of H.264/AVC video streaming is reached by introducing such network aware cross-layer algorithms in the UTRAN (UMTS Terrestrial Radio Access Network).