Doctor's Theses (authored and supervised):
"Non-Stationarity in Vehicular Wireless Channels";
Supervisor, Reviewer: C. Mecklenbräuker, C. Oestges;
oral examination: 04-23-2012.
n vehicular communications, the scattering environment changes rapidly. The fad- ing process in these channels is time-, and frequency-selective, and its statistical properties do not remain constant (stationary) for infinite time and infinite band- width. The stationarity in the time domain is defined by the validity of the wide sense stationary (WSS) assumption, and the stationarity in the frequency domain by the validity of the uncorrelated scattering (US) assumption. Hence, the fading process in vehicular channels is non-stationary, i.e. non-WSSUS. However, a non- stationary process can be divided into consecutive stationarity regions with finite extension in time and frequency where the WSS and US assumptions are valid, allowing to calculate its statistical moments.
In this thesis, I estimate the time- and frequency-varying scattering function of an observed non-WSSUS fading process. For that, I use the multi-taper based local scattering function (LSF) estimator. I use radio channel measurements collected in the DRIVEWAY´09 measurement campaign. DRIVEWAY´09 focuses on scenarios for intelligent transportation systems. First, I define a minimum stationarity region, and I present an optimal parametrization of the LSF estimator. For that purpose, I use the structure of a two-dimensional Wiener filter and optimize the parameters of the estimator to obtain a low mean square error (MSE) at the output without implying a large computational complexity for the estimation of the LSF. I show that there is an optimal combination of the estimator parameters which provides a good trade-off between MSE and computational complexity.
I also investigate the extension of the minimum stationarity region. I propose to use two spectral metrics to be applied to neighboring LSFs in order to perform a WSS and a US test: the spectral divergence and the collinearity. I prove that the fading process in vehicular communications is strongly non-WSS. Furthermore, for the first time (to the authors´ knowledge), I show their non-US behavior as well, mainly in scenarios with rich scattering. The dimensions of the minimum stationarity region are in the order of 40 ms in time and 40 MHz in frequency.
I stochastically characterize the time-varying vehicular channel parameters in terms of its first order moment (K-factor), and its second order moment (root mean square (RMS) delay and Doppler spread). I fit the obtained data to a bi-modal Gaussian mixture distribution, which is adequate for describing the channel parameters when both line of sight (LOS) and non-line of sight (LOS) conditions occur.
The small-scale fading of the envelope of the first delay bin is Ricean distributed with a varying K-factor. The later delay bins are mostly Rayleigh distributed. I demonstrate that the K-factor can not be assumed to be constant in time, frequency, and space. I show that the frequency-varying antenna radiation patterns as well as the time-varying number of active scatterers are the cause of these variations.
I also analyze the time-varying RMS delay and Doppler spreads, derived from the LSF. High RMS delay spread values are observed in situations with rich scattering, while high RMS Doppler spreads are obtained in drive-by scenarios.
Finally, I characterize the scattering environment by using a clustering algorithm to group multipath components stemming from the same scatterer. I apply the clustering algorithm on the LSF and calculate the time-varying cluster parameters. The cluster shape and the number of clusters depend on the richness of the scattering environment and the velocities of the scatterers. The cluster with the shortest delay has larger extension in comparison with later clusters.
vehicular communications, radio channel characterization, OFDM, IEEE 802.11p, MIMO, intelligent transport systems
Electronic version of the publication:
Created from the Publication Database of the Vienna University of Technology.