Variability and Confidence Intervals of the Power Measured in a Reverberation Chamber

Abstract

International audienceThe relative variance (or variability) of the average power measured in a reverberation chamber is derived. It is found to be a function of the average number of modes overlapping in the average modal bandwidth found at the working frequency. The model can predict the average-power variance from under- to over-moded regimes. Good agreement with experimental and simulation data is obtained. Confidence intervals of the estimate of the average-power variability are computed for different chamber regimes and a varying number of independent stirrer positions

Time-Reversed Excitation of Reverberation Chambers: Improving Efficiency and Reliability in the Generation of Radiated Stress

“…In order to answer this question, two points should be considered: the random nature of W (ν) and the fact that, in case of harmonic excitations, an RC is typically operated jointly with a stirring technique. The assumption of an overmoded RC, as defined in [10], implies that W (ν) is bound to follow an exponential probability law [11], [12]. As a result, η CW , when computed for a single realization, is widely dispersed around its mean value: e.g., η CW / η CW ∈ (0.05, 3.00) with 90 % probability.…”

A Possible Minimum Relevance Requirement for a Statistical Approach in a Reverberation Chamber

“…1. The sketch highlights indeed that only a finite number of fictitious modes intervene in (5). It follows that E e (r, f) can be simplified and reads…”

“…The real and imaginary parts ofγ i are assumed to follow a centered normal law. As done in [5], [8], and [9], we consider three sets of modal parameters, i.e.,γ i , ψ i , andξ i as random variables. The main assumptions required are at two levels.…”

“…In other terms, (2) and (6) must lead to the same expression of the variability. This one was shown to be expressed in [5] as…”

“…The relative variance of the measured mean power tends to unity, as P follows a χ 2 2 law. A recent study [5] has highlighted the rate at which power variability tends toward its asymptotic value.…”