Exploratory Data Analysis on Stochastic Emissions Near-Field Scanning Measurements

Recently, near-field scanning techniques for measuring and analysing stochastic emissions have been theoretically and experimentally studied and they are promising in terms of the characterization and classification radiated emissions in densely integrated technologies. Nonetheless, one of their drawbacks is the volume of data generated by time-domain acquisitions made across the scanning grid. Moreover, there are a multiplicity of sources of systematic errors that can pass unnoticed in measurements and that could compromise the accuracy of the data used for post-processing. This paper presents a pre-evaluation procedure that follows an exploratory data analysis approach for detecting such errors using an objective statistical criterion when applied to stochastic emissions of near-field scanning measurement data. Results are presented using real measurement data from the scanning of an Arduino Intel Galileo board, as has been reported in previous studies. By performing the proposed analysis, significant errors were detected in the measurement data. Final remarks on the importance of performing quality clearance in the measurement data from stochastic emissions assessments are provided.