Calibration of reverberation chamber for wireless measurements: Study of accuracy and characterization of the number of independent samples

Traditionally, the Reverberation Chamber (RC) has been used for EMC testing, but recently it has been developed for measuring antennas and mobile terminals designed for wireless applications. It is a metal cavity, large in terms of wavelength which supports several cavity modes at the operation frequency. The modes are stirred by appropriate stirring methods producing a rich scattering environment with Rayleigh distribution of the received complex signal amplitude. Due to the statistical nature of the electromagnetic field inside the chamber, the RC is ideal for simulating in compact space the behavior of mobile terminals operating in fading environments as the urban and the indoor ones. The rapidly growing use of the RC for these kinds of applications has developed the necessity of fast and accurate measurements. In addition, the accuracy is typically more critical in antenna efficiency measurements than in EMC tests. Therefore, studies of the mechanisms that may affect the measurements accuracy are becoming of great interest. If the stirrers are able to perform a proper stirring of the excited modes, a large number of independent field samples are created. The independency of those samples is of great importance in quantifying the uncertainty of a test performed in the cavity. This thesis determines the importance of different steps in the procedure of calibrating a RC in order to improve the accuracy for wireless measurements. The study is performed using the Bluetest High Performance RC at Chalmers University of Technology. By measuring with several stirring configurations, number of stirrer positions and chamber loading, we are able to show how the number of independent samples varies when different stirring techniques are used and extra lossy objects are present inside the chamber. In addition, a new theoretical formula for the estimation of the number of independent samples when different stirring methods are applied is proposed. The results show that the number of independent samples estimated with the new formula and the one measured have a poor agreement at the lower frequencies compared to high fre-