IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 58, NO. 2, APRIL 2016 385 Decomposition of Electromagnetic Interferences in the Time-Domain Marco A. Azp ´ urua, Member, IEEE, Marc Pous, and Ferran Silva, Member, IEEE Abstract—Electromagnetic interferences are potentially very complex signals formed by the superposition of transient (broad- band) and continuous wave (narrowband) components with signif- icant randomness in both amplitude and phase. Decomposing the electromagnetic interference measured in the time domain into a set of intrinsic mode functions is useful to gain insights of the pro- cess that generates the interference. Evaluating the intrinsic mode functions contributes to improving the measurement capabilities of the time-domain electromagnetic emissions measurement sys- tems based on the general-purpose oscilloscopes. In this paper, a combination of techniques that includes empirical mode decom- position and transient mode decomposition is used to separate the main components of complex electromagnetic disturbances. This approach requires no prior information on the spectral content of the measured EMI and it does not perform a domain transfor- mation. Examples of electromagnetic interference decomposition verify the effectiveness and the accuracy of the proposed approach. Finally, a discussion on the advantages, practical applications, lim- itations, and drawbacks of the described techniques is addressed. Index Terms—Digital signal processing, electromagnetic com- patibility, electromagnetic interference, electromagnetic measure- ments, time-domain analysis. I. INTRODUCTION E LECTROMAGNETIC interferences are complex signals composed of the superposition of continuous-wave, tran- sient, and random disturbances. In consequence, measuring properly and defining specific features of an electromagnetic interference (EMI) are challenging tasks. In this regard, the conventional approach for the evaluation of the electromagnetic emissions consists in measuring the am- plitude spectrum of the disturbance using a frequency sweep receiver and the CISPR standard detectors [1]. Therefore, the current standard measurement procedures for assessing an EMI neglect important time-domain characteristics of the electro- magnetic disturbances, such as the repetition rate, the different emission profiles of an EUT as it changes its operation mode, and the impact of transient events [2]. This is particularly relevant for evaluating and predicting the degradation suffered by digital Manuscript received June 16, 2015; revised November 20, 2015; accepted January 4, 2016. Date of publication January 25, 2016; date of current ver- sion March 8, 2016. This work was supported in part by the EURAMET IND60EMC research project (the EMRP is jointly funded by the EMRP par- ticipating countries within EURAMET and the European Union) and by the Spanish “Ministerio de Econom´ ıa y Competitividad,” under project TEC2013- 48414-C3-3-R. The authors are with Group of Electromagnetic Compatibility, Department of Electronic Engineering, Polytechnic University of Catalonia, Barcelona 08034, Spain (e-mail: marco.azpurua@upc.edu; marc.pous@upc.edu; ferran.silva@ upc.edu). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TEMC.2016.2518302 communication systems due to the influence of a transient EMI [2]. Consequently, in some cases, the standardized methods of EMI measurement and evaluation are not completely suitable, insufficient or might require an enormous amount of time to provide reliable measurement results [3], [4]. Currently, the fast Fourier transform-based EMI test re- ceivers and real-time spectrum analyzers have time-scan mea- surement capabilities that overcome many of the limitations of the stepped frequency scan in the EMI receivers. However, the bandwidth of their intermediate frequency filter and the tradeoff between the time domain and the frequency-domain resolutions [5] are their main constraints for measuring simulta- neously broadband and narrowband interferences on the whole frequency range. Recently, advances in an EMI postprocessing and the enhanced capabilities of digital oscilloscopes allowed to implement entirely time-domain EMI (TDEMI) measure- ment systems. Those TDEMI systems provide accurate, full spectrum, multichannel, time saving, and cost-effective EMI measurements [6]. Nonetheless, estimating the spectral content of an EMI is only one side of a multifaceted problem. For example, tran- sient, intermittent, or event triggered (i.e., by changing opera- tion modes) disturbances could be unnoticed by conventional frequency sweep EMI measurements because the unsynchro- nized occurrence of the EMI and its short duration. Previous research has addressed some of those problems using time- domain EMI analysis. In particular, Alban et al. [7] used an ad hoc oscilloscope-based TDEMI measurement system for devel- oping a statistical model for the broadband noise in comput- ing platforms. In the field of power electronics, time-domain EMI measurements have also been used for providing time- frequency-energy distributions by means of the wavelet analysis used for emissions mitigation in chaotic converters [8]. More re- cently, statistical approaches to the evaluation of EMI measured in the time-domain have been used to predict the degradation caused by transient disturbances on digital communication sys- tems [2], [9] and to model the distribution of peak measurements using extreme value theory [10]. In that sense, exploring new processing approaches and tech- niques is fundamental to enhance the performance of such TDEMI systems. In particular, this paper presents an entirely time-domain method that comprises techniques and algorithms for signal decomposition that offer further possibilities to EMI analysis. This method is based on the empirical mode decompo- sition (EMD) [11] and includes a preliminary stage of transient EMI separation that improves EMD performance. The decomposition of an EMI in the time-domain seeks to decompose the main components of the measured signal 0018-9375 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.