RF-Driven Crowd-Size Classif i cation via Machine Learning

In this letter, we propose a machine learning solution for crowd-size classification in an indoor environment. Narrow-band radio frequency signals are used to identify a pattern according to the number of people. Experimental data collected by a low-cost software-defined radio platform are postprocessed by applying a feature mapping along with the random forest technique for classifying the crowd-size scenarios. The proposed solution has significant accuracy in classification performance.