Measurement of interferometric parameters values is affected by phase disturbance due especially to atmospheric turbulences. Algorithms of fringe sensing, aimed at fringe parameters identification, are based on interferometric models that have to be carefully adapted to the interfered beams, including variability sources. All information is contained in the collected signals, and how to extract it is subject of researches. In the first part of the thesis, we present the fringe sensing algorithms proposed for two different fringe sensors for the ESO VLTI: FINITO and PRIMA FSU. We highlight how they must adapt to the different instrumental layouts, and we summarize their performance both on simulated and on laboratory data. In the second part, we show the results of the application of some statistical techniques to real interferometric signals. With the time series analysis we examine the composition of signals before and after the combination, separating pure random effects and peculiar features from trends. With the multivariate regression analysis, we are able to isolate noise components due to the interference physical process. These techniques can be useful for quality tests both for instrumental performances and for signals characterization. They could be easily adapted to different interferometric instruments.