Análise comparativa entre diferentes frequências de coleta de dados geolocalizados.

Abstract

Geolocation data has been widely used for the comprehension of various social phenomena. Nowadays, such data are produced at large scale by people, especially by their smartphones equipped with various types of sensors. However, the capture os this kind of data, in a mobile device, can be expensive as the collection rate increases and may consume device resources. On the other hand, the capture with low frequency may impair the quality and consistency of the information collected. In this context, we conducted a comparative of performance across different data collection frequencies (of 15, 30, 60 and 120 seconds) to analyze the impact on resource consumption and data quality. For this study, we used an application developed in a previous work [1]. This app is able to collect geolocated data from several sensors embedded in the smartphone. The experiment had 10 volunteers, who used the application for a period of 20 days. Afterwards, an evaluation was performed taking into account aspects related to the device resource consumption and aspects regarding the quality of the data captured, in order to show the pros and cons of the different capture frequencies and estimate a frequency best suited for different usage scenarios. As a result, the 30-second frequency collection had the best tradeoff between resource consumption and information generation, and the 60-second frequency collection was the only one that showed no better effectiveness in any specific application compared to the other frequencies.