Design of Mobile Radio Channel Simulators Using the Iterative Nonlinear Least Square Approximation Method with Applications in Vehicle-to-X Communications

Doktorgradsavhandling i informasjons- og kommunikasjonsteknologi, Universitetet i Agder, 2015Vehicle-to-X (V2X) communication systems are expected to provide tremendous benefits associated with the safety and traffic efficiency on roads. The successful deployment of emerging technologies like V2X requires channel models accurately representing fading statistics in environments where those technologies are used. The accuracy is, of course, a major concern when adapting or developing a suitable channel model for test and evaluation purposes. However, it is also important to take into account the simplicity of a channel model, which is crucial for efficient numerical computations and computer simulations. Reconciling simplicity and accuracy is a rather complex task to accomplish, which requires sophisticated parameter computation methods. To the best of our knowledge, only a limited number of investigations address the channel modelling and parametrization problems for vehicular propagation scenarios in the literature. In order to fill this gap, we concentrate on the development of new sophisticated channel modelling approaches and efficient parameter computation methods for the design of V2X communication systems in this dissertation. In general, there are two main applications of channel models: (1) for the design and test of wireless communication systems and (2) for the optimization of existing communication systems. For the design and test purposes, more general statistical models such as Rice and Rayleigh channel models are preferred. Those channel models provide a fundamental insight into propagation phenomena and at the same time they greatly simplify the theoretical and numerical computations to assess the performance of wireless communication systems. For the optimization purposes, however, measurement-based channel models are commonly used. The main advantage of such channel models is that they always accurately reflect the physical reality. In this dissertation, we will focus on the channel models designed for both of those application purposes. A significant part of this dissertation will be devoted to the thorough analysis and design of Rayleigh and Rice fading channel models. We investigate the correlation properties of those channels assuming asymmetrical shapes of Doppler power spectral densities (PSDs). In fact, this is what we often observe in real-world propagation scenarios. In this regard, we will present an analytical expression for the autocorrelation function (ACF) of Rice processes that captures such realistic scenarios. Another important contribution to this topic is the novel iterative nonlinear least square approximation method for the design of Rice and Rayleigh channel simulators based on sum-of-sinusoids (SOS), as well as sum-of-cisoids (SOC) approaches. The idea behind the proposed method is very simple. The parameters of the simulation model are extracted from the reference model, such as the stochastic Rice and Rayleigh channel models, by fitting the statistical properties of interest, e.g. the ACF and the probability density function (PDF). We show that the proposed method outperforms several other methods in designing channel simulators with desired distribution and correlation properties. We also show that the proposed method provides a subtle balance between channel model’s simplicity and accuracy in designing Rayleigh and Rice channel simulators. The parametrization is a process of determining the key parameters specifying the channel model. This process has a great influence on the reliability of the developed channel model. It is therefore highly desirable if those parameters are extracted from measurements. In fact, this idea constitutes the fundamental concept behind measurement-based channel modelling approach. The measurement-based models are important in the sense that they can be used for the optimizations of the wireless communication system. Hence, the problem of computing the channel model parameters from the measurements is of special interest. In this regard, we propose iterative nonlinear least square approximation method for the design of measurementbased channel simulators. Through detailed investigations and comparative studies, we demonstrate that the proposed method is highly flexible and outperforms several other conventional methods in terms of reproducing the correlation characteristics obtained from several measurements. In addition, we introduce a new approach for the design of channel models for V2X communications in tunnel environments, where the number of scatterers contributing to the total received power is relatively small