Performance Evaluation of Real-Time Multivariate Data Reduction Models for Adaptive-Threshold in Wireless Sensor Networks

This article presents a new metric to assess the performance of different multivariate data reduction models in wireless sensor networks. The proposed metric is called updating frequency metric that is defined as the frequency of updating the model reference parameters during data collection. A method for estimating the error threshold value during the training phase is also suggested. The proposed threshold of error is used to update the model reference parameters when it is necessary. Numerical analysis and simulation results show that the proposed metric validates its effectiveness in the performance of multivariate data reduction models in terms of the sensor node energy consumption. The adaptive threshold improves the frequency of updating the parameters by 80% and 52%, in comparison to the nonadaptive threshold for multivariate data reduction models of MLR-B and PCA-B, respectively.