Universidade Federal do Ceará
Mestrado e Doutorado em Ciências da Computação

Título: ALOG: Adaptive Longitudinal Grids for Geospatial Data using Local Differential Privacy

Data: 13/02/2025
Horário: 08h30
Local: Sala de Seminários – Bloco 952

Resumo:

This study introduces ALOG (Adaptive Longitudinal Grids for Geospatial Data using Local Differential Privacy), a novel framework designed to optimize geospatial data collection and frequency estimation while ensuring robust user privacy. ALOG leverages adaptive grids to dynamically adjust spatial granularity based on data density, eliminating the need for prior knowledge about data distribution. Three ALOG variations are evaluated: ALOG-2R, which applies a dual-round sanitization process, and two single-round adaptations, ALOG-1R-a and ALOG-1R-b. These approaches are compared with existing protocols, such as RAPPOR, LOLOHA, and L-OSUE, prior state of the art, using both synthetic and real-world datasets. Experimental results demonstrate ALOG’s superior performance in balancing privacy and utility, particularly under varying grid sizes and privacy budgets. The findings highlight the effectiveness of adaptive grid refinement in achieving precise frequency estimates in privacy-sensitive applications without relying on prior knowledge of data density. ALOG’s flexible design represents a significant advancement in privacy-preserving geospatial data analysis, opening the door to broader applications in location-based services.

Banca examinadora:

• Prof. Dr. Javam de Castro Machado (MDCC/UFC) – Orientador
• Prof. Dr. José Maria da Silva Monteiro Filho (MDCC/UFC)
• Prof. Dr. Sérgio Lifschitz (PUC-Rio)
• Prof. Dr. Felipe Brito Timbó (LSBD)