Characterization of Single- and Multi-antenna Wireless Channels

Abstract: The wireless propagation channel significantly influences the received signal, so that it needs to be modeled effectively. Extensive measurements and analysis are required for investigating the validity of theoretical models and postulating new models based on measurements. Such measurements, analysis, and modeling are the topic of this thesis. The focus of the included contributions are Multiple-Input Multiple-Output (MIMO) propagation channels and radio channels for sensor network applications. Paper I presents results from one of the first MIMO measurements for a double-directional characterization of the outdoor-to-indoor wireless propagation channel. Such channels are of interest for both cellular and wireless LAN applications. We discuss physical aspects of building penetration, and also provide statistics of angle and delay spreads in the channel. The paper also investigates the coupling between DOD and DOA and the two spectra are found to have non-negligible dependence. We test the applicability of three analytical channel models that make different assumptions on the coupling between DODs and DOAs. Our results indicate that analytical models, that impose fewer restrictions on the DOD to DOA coupling, should be used preferrably over models such as the Kronecker model that have more restrictive assumptions. Paper II presents a cluster-based analysis of the outdoor-to-indoor MIMO measurements analyzed in Paper I. A subset of parameters of the COST 273 channel model, a generic model for MIMO propagation channels, are characterized for the outdoor-to-indoor scenario. MPC parameters are extracted at each measured location using a high-resolution algorithm and clusters of MPCs are identified with an automated clustering approach. In particular, the adopted clustering approach requires that all MPC parameters must be similar in order for the MPCs to form a cluster. A statistical analysis of the identified clusters is performed for both the intra- and inter-cluster properties. Paper III analyzes the spatial fading distribution for a range of canonical sensor deployment scenarios. The presented results are relevant to communicating within, and between, clusters of nodes. Contrary to the widely accepted assumption in published literature that the channel is AWGN at a small-enough distance, our measurements indicate that values of the Rice factor do not, in general, increase monotonically as the Tx-Rx distance is reduced. A probability mixture model is presented, with distance dependent parameters, to account for the distance dependent variations of the Rice factor. A simulation model that includes small- and large-scale fading effects is presented. According to the modeling approach, a sensor node placed anywhere within the spatial extent of a small-scale region will experience the channel statistics applicable to that region. Paper IV presents results characterizing a radio channel for outdoor short-range sensor networks. A number of antennas are placed on the ground in an open area and time-variation of the channel is induced by a person moving in the vicinity of the nodes. The channel statistics of both the LOS path and the overall narrowband signal are non-stationary. We investigate the stationarity interval length to be used for small-scale analysis. Our analysis of the various measured links shows that the Rx signal strength is significantly influenced by a moving person only when the person blocks the LOS path. We present a generic approach for modeling the LOS blockage, and also model the time-variant Doppler spectrum of the channel's scattered components.