Space-time parameter estimation and statistical modeling of the indoor radio channel

In the past few years it has been proposed to incorporate adaptive antenna subsystems into wireless system designs, as a method to achieve enhanced performance, and provide new features such as geolocation. An adaptive, or smart antenna, provides these benefits by applying space-time processing to the signals delivered by an array of antenna elements. As with any wireless system, the performance achieved by a smart antenna is fundamentally dictated by the properties of the propagation environment in which it operates. While a large collection of channel models is available to support the design of a wireless system employing omnidirectional elements, essentially no validated models are available which are suitable for the assessment of a smart antenna system.; Analyzing and optimizing the performance of an adaptive antenna system requires a channel model based upon a joint characterization of the time, angle and complex amplitude with which paths arrive at the receiver. This work presents an investigation into the problem of identifying and modeling these properties as they occur in the indoor radio environment. Many of the existing algorithms for array signal processing are reviewed and analyzed, and shown to exhibit a high degree of sensitivity to errors in the underlying signal model assumptions, such as the form of the array manifold, and the correlation properties among the impinging signals. A theoretical framework is derived that relates the minimum achievable channel identification error to the fundamental parameters of the measurement system. In both time-only and joint space-time measurement, many of these parameters are conveniently adjusted to provide channel identification of arbitrarily high quality. The design and calibration of a data acquisition system based upon a circular antenna array, and frequency domain excitation, are presented. As this array does not display shift-invariant signal characteristics, two new algorithms are derived that exploit knowledge of calibration data to provide accurate estimates of the space-time channel impulse response. The performance of these algorithms, both when processing a large number of simultaneous arrivals and at low received signal-to-noise ratios, is compared by operating upon synthetic data sets. A field study is described in which 500 data sets were acquired, together with highly-accurate physical measurements. Two new statistical models are derived from these data sets, and compared with the recently-proposed Geometric-Based Single-Bounce Elliptical and Extended Saleh-Valenzuela models. The new models are shown to provide the closest agreement with the measured results.