A Wide-Band Air-Ground Channel Model

Kurzfassung

In civil aviation, air-traffic management and with that the underlying communication, navigation, and surveillance (CNS) infrastructure is currently being modernized. To test improved or newly developed terrestrial CNS systems without costly implementations and measurements, computer simulations using an AG channel model are crucial. In this dissertation we propose an AG channel model for the L-band which unlike previous proposals - fully captures the statistical non-stationary nature of the channel: we are now able to test the full range of terrestrial CNS systems. We propose a geometry-based stochastic channel model (GBSCM) architecture for the AG channel. In the GBSCM, geometrical model elements represent the different propagation effects, e.g., reflections off buildings. The model elements and their properties are environment specific and are characterized by statistical distributions: the channel model can be adapted to different environments by choosing the respective statistical distributions. We conducted two L-band flight trial campaigns in a regional airport environment to investigate the AG channel. Throughout this dissertation we use the collected data to parameterize the channel model architecture. To characterize the geometrical model elements for the parameterization, multipath components (MPCs) in the measured signal need to be detected and their parameters have to be estimated: we propose an algorithm combining super-resolution parameter estimation with a tracking filter. Based on the estimated MPC parameters we propose localizing the reflectors causing these MPCs using a Bayesian method. The presented algorithms may not only be used in AG channel sounding but in any scenario where an accurate MPC parameter tracking and reflector localization is required. Applying the algorithm to the measurement data demonstrates the non-stationarity of the AG channel. The results also allow us to parameterize the channel model to a regional airport environment. To assess the channel model quality the parameterization is successfully validated against measurement data. In consequence, our newly developed channel model is of high value for the computer-based performance analysis of CNS systems: the performance of new or improved CNS systems can now be realistically tested using computer simulations.