Redes multicamadas para cidades inteligentes: estudo, modelagem e utilização

Abstract

This thesis presents an in-depth analysis of urban mobility in Curitiba and disease prevention strategies in public transport systems, using multilayer network models. The research begins by characterizing public transport mobility patterns by economic classes in Curitiba. Using a smart card dataset, the population was segmented into economic strata to analyze public transport usage patterns in space-time, such as departure times and destinations reached by different classes. An origin-destination network was built to reveal distinct mobility patterns, showing that, with increasing income, morning peak hours are, on average, delayed by 2 hours, and the spatial distribution of trips becomes more localized compared to lower classes. Validation of the results with an official household mobility survey in Curitiba demonstrated that the proposed model accurately captures realistic mobility patterns, using a more accessible and cost-effective data source. The study concludes that the difference in mobility patterns between classes is noticeable, despite subtle income differences, and that this understanding can subsidize proposals for more sustainable economic development and urban interventions in the public transport system. Furthermore, the study focuses on strategies to reduce contamination in public transport using a multilayer network approach. The thesis proposes the identification of critical locations for prioritization in interventions, such as vaccination campaigns, in pandemic scenarios, considering real urban traffic data. Two Multi-Aspect Graph (MAG) models were developed: MAG 1, with four aspects and two layers (smart card users and bus stops), and MAG 2, with three aspects and three layers (health units, identifier, and neighborhood). The Percolation Centrality metric was employed to quantify the relative importance of disease dissemination at each vertex of the network. The flexibility of the approach was also evidenced by the identification of critical health units for targeted vaccination. The study highlights that the approach can be applied to other problems, such as public transport optimization, and that the methodology offers valuable insights for public policymakers to optimize resource allocation in disease prevention campaigns and urban planning.