COMPLETE ENSEMBLE EMPIRICAL MODE DECOMPOSITION (CEEMD) FOR REAL-TIME SIGNAL DETRENDING IN IOT APPLICATIONS

The Internet of Things (IOT) is a promising area which will boost the world economy. The
constituent components of the IOT are smart objects which generate actuation signals or receive
sensory signals which are usually noisy, have trend or has small signal-to-noise ratio. Processing these
signals for filtering, detrending and enhancing the signal-to-noise ratio is crucial for embedding
intelligence in these smart objects. This research discovers the potential of CEEMD in preparing
signals for further intelligent applications such as event detection or pattern recognition in smart
objects. Algorithms are presented for signal filtering, detrending and event detection based on a
combination of both CEEMD, the autocorrelation function and the learning vector quantization
classifier.The performance of the proposed algorithms is compared for both CEEMD and the least
squares fit approach. The CEEMD has shown promising results.