Hierarchical clustering-based assessment of telephone-audible noise: time- and frequency-domains-based approach

This study presents a hierarchical clustering-based assessment of audible noises at telephone terminals. In discriminating the electromagnetic interference (EMI) patterns of telephone-audible noises, both the time- and frequency-domains-based approaches are applied. Here, signal occupation ratios for several amplitude intervals (such as from -0.1 to 0.1, from -0.2 to 0.2, etc. compared to the normalised maximum amplitude) of the noises were calculated for time-domain assessment; compositions of multiple energy spectra for divided frequency bands (such as 1-100 Hz, 100-200 Hz, etc.) of the noises were derived for frequency-domain assessment. In evaluating our method, the authors measured 12 kinds of telephone-audible noises (e.g. due to radio waves etc.) and then input their properties into the hierarchical clustering scheme. As a result, the authors visually confirmed that measured telephone-audible noise data sets could be classified into several groups for both cases. The result for time-domain assessment clearly reflected EMI effects of each EMI source, while frequency-domain assessment tended to classify data sets more clearly according to the kind of EMI source. The proposed method can deal with multi-dimensional parameters, projecting its results onto a tree-like structure in which data categories give us an effective depiction of EMI sources that cause disturbances at telephone terminals.