Wearable Bio-Signal(PPG)-Based Personal Authentication Method Using Random Forest and Period Setting Considering the Feature of PPG Signals

Abstract

A study regarding personal authentication based on PPG has been conducted using the Random Forest algorithm. In order to ensure correct authentication, data features must be consistent. This consistency is provided through the normalization of the PPG signal using maximum-minimum normalization and spline interpolation. The threshold is set by using normalized data and the highest value among the points above the threshold is set as the peak of the period. After establishing candidates of valley with values below the threshold, a shifted period is calculated by designating the last valley next to the following peak to the valley of the period. In order to reduce errors in the one period PPG data, the data is averaged after overlapping. A discrete cosine transform (DCT) is used to extract features from the preprocessed data. Thus, the extracted features are used as the input variables for machine learning techniques such as Decision Tree, KNN, Random Forest. The accuracy of these algorithms is 93%, 98%, and 99% respectively.

Photoplethysmography Biometric Recognition Model Based on Sparse Softmax Vector and k-Nearest Neighbor

“…Here, the training template and test sample generated by the proposed feature-extraction method are input into the machine-learning classifier, and then the recognition rate is output. e classifiers used here include k-NN [4,10,25], NB [7,26], RF [13,27], and LDC [3].…”

“…. , 16} do (11) let S j v � GNF X j (y j , w, K) (12) end for (13) the proposed method achieves recognition rates of 99.73%, 99.78%, and 99.92% for one, two, and three layers based on k-NN on the CapnoBase dataset, respectively. In short, all these results demonstrate the effectiveness of the three-layer feature extraction method.…”

“…Many of the previous methods focused on fiducial approaches, but fiducial-point detection in any condition is error prone. erefore, many researchers [9][10][11][12][13] have researched nonfiducial-point-based approaches. Nonfiducial approaches take a more holistic approach in which features are extracted statistically based on the overall signal morphology [10].…”

“…Farago et al [12] presented the correlationbased nonfiducial features extraction technique and achieved a 98% accuracy rate for 30 subjects by computing correlations of the individual's peak-to-peak interval to a reference PPG peak-to-peak interval. Lee et al [13] tried to use a discrete cosine transform (DCT) for extracting features from the preprocessed PPG data, and the extracted features were used as the input variables for machine-learning techniques, e.g., decision tree, k-NN, and random forest (RF); the accuracy of these aforementioned algorithms was 93%, 98%, and 99%, respectively.…”

Transcending Conventional Biometry Frontiers: Diffusive Dynamics PPG Biometry

“…Focusing now on the matter at hand, the first documented reference to the PPG-based biometric system dates back to Gu et al’s research in 2003 [ 42 ]. In all the works that use the PPG signal as a biometric reference, specific biomarkers correspond to features implicitly or explicitly extracted from the signal waveform: for example, time-domain features acquired from the PPG signal’s first and second derivatives for biometric identification [ 43 ], approximating each PPG signal as a sum of Gaussians and using the parameters in a discriminant analysis framework to distinguish individuals [ 44 ], or defining the waveform of the PPG signal in five consecutive PPG cycles [ 45 ], from 22 cycles [ 46 ] or 100 cycles [ 47 ] parametrically. One of the latest works is related to the non-fiducial and fiducial approaches for feature extraction with supervised and unsupervised machine learning classification techniques [ 48 ], recently expanded with other multi-feature classification techniques [ 49 , 50 ].…”