Nonparametric bootstrap estimation of confidence interval in base station test

In most test analysis, we always want to construct a parameter model, and then through sample statistics and model parameters, we can get the properties, or characteristics of what we concern about. Confidence intervals can give an estimate of the range within which the true value of the statistic lies. And a narrow confidence interval indicates the low variability of the statistic, which can give a strong support for the conclusion made from the statistical analysis. In base station test, we can barely construct an accurate parameter model, because the measured value varies with many factors. Without theoretical formulas, we can not get accurate assessment of the measured value. The Efron bootstrap statistical analysis can just solve this problem. In this article, we introduce several nonparametric bootstrap methods in assessing the accuracy of sample statistic, and the validations of these methods are performed with both Measured data and simulation data. We have found that all of those methods give an accurate prediction of the 90 percent confidence interval for the mean. And they can still work even under small data sets.