TY - JOUR
TI - Channel modeling approaches to wireless system design and analysis
DO - https://doi.org/doi:10.7282/T3959H98
PY - 2010
AB - In wireless communications, it is common practice to use mathematical models for describing the radio channel. One approach is stochastic modeling, in which the key properties of the signal propagation (e.g., multipath fading) are captured by probability distributions. If the interest pertains to a specific environment, an alternative approach is to measure channel responses for a very large population of transmit-receive (T-R) paths; this is an effective but labor-intensive approach. An alternative approach that is less costly and more flexible is to use environment simulators. These are computer programs that (1) emulate the physical environment; (2) use wave propagation physics to predict the radio signal produced at any receive point from any transmit point; and (3) account for transmission through walls and diffraction around walls. This works best when the user has site-specific information on the geometry and structure materials When the physical environment is well-specified, such as indoor areas where the layouts and materials of walls, floors and ceilings are known, environment simulation can be employed on a very large scale with very little effort. In this thesis we focus on environment simulators based on ray-tracing. The major contribution is to demonstrate and evaluate the use of ray tracing for characterizing wireless channels and analyzing algorithms for various applications. We initially demonstrate, via comparisons with physical measurements, the statistical accuracy of ray-tracing predictions of channel behavior. The comparisons are made for three parameters that largely characterize a radio path's behavior: Path loss; Ricean K-factor; and RMS delay spread. The comparisons show good agreement over the set of paths measured and simulated, establishing confidence that a well-designed radio simulator can be used reliably in system studies. Environment-specific models generally assume the channel response is non-varying over time if both ends of the path are fixed. However, in real environments, channel responses vary over time, e.g. due to movement of objects (or people) near the transmission path. We have measured the channel response in an office building under different scenarios of environment dynamics. We stochastically modeled the time variation of the channel response about the mean using autoregressive processes and showed that this can lead to an accurate representation. Our approach could be used to model the time-varying tap gains to further augment the realism of ray-tracing simulations. We then demonstrate several applications in wireless system design where ray-tracing could be exploited. First, we present an algorithm called Emitter Localization and Visualization (ELVIS) for localizing emitters by back-propagating the received signals via back-ray tracing. Second, we present a statistical path loss model derived from data simulated using a ray tracing tool. The characterization used is a nonlinear curve of the dB path loss to the log-distance, with a random variation about that curve due to shadow fading. Third, we devise an evaluation approach for densely populated urban wireless systems using MIMO links, wherein the location-specific channel gains are determined via ray-tracing. We compare and quantify the data rate performances of MIMO systems for various transmission schemes and antenna configurations; we present algorithms for adapting the MIMO transmission mode to varying channel conditions as a mobile moves along a given trajectory; and we treat the case of multiple bases to cover a full urban neighborhood and investigate the relationship among frequency reuse, co-channel interference and achievable data rate.
KW - Electrical and Computer Engineering
KW - MIMO systems--Design
KW - Ray tracing algorithms
KW - Stochastic models
LA - eng
ER -