60 GHz Wireless Propagation Channels: Characterization, Modeling and Evaluation

Sammanfattning: To be able to connect wirelessly to the internet is nowadays a part of everyday life and the number of wireless devices accessing wireless networks worldwide are increasing rapidly. However, with the increasing number of wireless devices and applications and the amount available bandwidth, spectrum shortage is an issue. A promising way to increase the amount of available spectrum is to utilize frequency bands in the mm-wave range of 30-300 GHz that previously have not been used for typical consumer applications. The 60 GHz band has been pointed out as a good candidate for short range, high data rate communications, as the amount of available bandwidth is at least 5 GHz worldwide, with most countries having 7 GHz of bandwidth available in this band. This large bandwidth is expected to allow for wireless communication with bit rates up to 7 Gbit/s, which can be compared to the typical WLAN systems of today that typically provide bit rates up to 0.6 Gbit/s. However, the performance of any wireless system is highly dependent on the properties and characteristics of the wireless propagation channel. This thesis focuses on indoor short range wireless propagation channels in the 60 GHz band and contains a collection of papers that characterizes, models and evaluates different aspects that are directly related to the propagation channel properties. Paper I investigates the directional properties of the indoor 60 GHz wireless radio channel based on a set of indoor measurements in a conference room. In the paper, the signal pathways and propagation mechanisms for the strongest paths are identified. The results show that first and second order interactions account for the major contribution of the received power. The results also show that finer structures, such as ceiling lamps, can be significant interacting objects. Paper II presents a cluster-based double-directional channel model for 60 GHz indoor multiple-input multiple-output (MIMO) systems. This paper is a direct continuation of the results in paper I. The model supports arbitrary antenna elements and array configurations and is validated against measurement data. The validation shows that the channel model is able to efficiently reproduce the statistical properties of the measured channels. The presented channel model is also compared with the 60 GHz channel models developed for the industry standards IEEE802.15.3c and IEEE802.11ad. Paper III characterizes the effect of shadowing due to humans and other objects. Measurements of the shadowing gain for human legs, metallic sheets, as well as metallic and plastic cylinders are presented. It is shown that the shadowing gain of these objects are fairly similar and that the shadowing due to the metal cylinder can be determined based on the geometrical theory of diffraction. Next, the shadowing due to a water-filled human body phantom is compared with the shadowing due to real humans. The results show that the water-filled phantom has shadowing properties similar to those of humans and is therefore suitable for use in 60 GHz human body shadowing measurements. Paper IV presents a novel way of estimating the cluster decay and fading. Previously, the cluster decay has usually been determined by performing a simple linear regression, without considering the effects of the noise floor and cluster fading. The paper presents an estimation method which takes these effects into account and jointly estimates both the cluster decay and cluster fading. It is shown that this estimation method can greatly improve the estimated parameters. Paper V evaluates the capacity improvement capability of spatial multiplexing and beamforming techniques for 60 GHz systems in an indoor environment. In this paper, beamforming refers to conventional gain focusing in the direction of the strongest propagation path. The paper uses a capacity metric that only depends on the multi-path richness of the propagation channel and the antenna aperture size. In the paper, it is shown that, when the link budget is limited due to electrically small antennas and long Tx-Rx separation distances, beamforming approximates the capacity of spatial multiplexing. However, spatial multiplexing is a worthwhile option when Rx SNR is favorable and a higher peak data rate is required. Paper VI describes different methods for the clustering of wireless multi-path components. In the literature, the clustering method that is predominantly used is the K-means algorithm, or a power-weighted version of K-means, called K-power means. In this paper, we point out that K-means is a special case of a Gaussian mixture model (GMM). The paper presents a clustering method based on a GMM. This method is able to handle arbitrary cluster spreads in the different dimensions better than the K-means algorithm. A power-weighted version of the GMM is also presented. In addition to this, a mixture model based on asymmetric Laplace distributions is also presented, with and without power-weighting. Paper VII is based on channel measurements in a small and a large room, where the Tx and Rx arrays have dual polarized elements. Using these measurements, the cross-polarization ratio (XPR) of the multi-path components are characterized. This gives valuable information on how the MPCs are affected by the propagation channel. A statistical description of the XPR is also needed for the development of a propagation channel model that supports polarization. The paper also investigates the eigenvalue spreads for single and dual polarized elements. Furthermore, the measurements include LOS and NLOS measurement, where the NLOS scenarios include water-filled human presented in paper III. The results show that the capacity can be greatly improved if dual-polarized elements are used, and that the XPR values are in general higher compared to lower frequencies.