An automated location detection method in multi-storey buildings using environmental sound classification based on a new center symmetric nonlinear pattern: CS-LBlock-Pat

Abstract

Classical navigations applications only give latitude and longitude information, thus, location detection has been processed in two-dimensional(2D) space by using latitude and longitude. There are many multi-storey buildings in cities, and these buildings cause location detection problem in 3D space. This research presents a solution for solving this problem. An environmental sound classification(ESC) dataset was collected from a multi-storey hospital, and an automated ESC model is presented. As a feature generation function, a new center symmetric nonlinear pattern is presented using a substitution box(S-Box) of LBlock lightweight block cipher., therefore, this model is named CS-LBlock-Pat. The presented model is applied to the dataset collected from a multi-storey hospital. This hospital has ten floors. Therefore, ten classes of classification results are given. This model yielded 95.38% accuracy rate using SVM. The results obtained from the classification phase obviously demonstrated that the 3D location detection problem could be solved by using ESC.