2026-01-292026-01-292025-03-31MALAQUIAS JUNIOR, Mozart Lima. Modelagem e predição de canal em 3,5 GHz com redes neurais: abordagens com simulações e medições em ambientes fechados. Orientadora: Jasmine Priscyla Leite Araújo. 2025. 75 f. Dissertação (Mestrado em Engenharia Elétrica) - Instituto de Tecnologia, Universidade Federal do Pará, Belém, ano de defesa. Disponível em: https://repositorio.ufpa.br/handle/2011/17903. Acesso em:.https://repositorio.ufpa.br/handle/2011/17903This work proposes a small-scale and large-scale channel analysis for the 3.5 GHz frequency in indoor environments, using the simulation software Wireless InSite (WI) and measurements carried out on the ground and first floors of the building annexed to the Electrical Engineering Laboratories at the Federal University of Pará. To analyze the correlation between measured and simulated data, the Root Mean Squared Error (RMSE) was used, with results of 3.681 dB for the ground floor and 6.059 dB for the first floor, validating the small-scale parameters obtained from the simulation. The Floating-Intercept and Close-In path loss models were applied to the measurement and simulation data. Based on these results, metrics such as the Power Delay Profile, mean delay, delay spread, and coherence bandwidth were determined. Additionally, a neural network model was proposed to predict the received power of rays arriving at each reception point, demonstrating the effectiveness of the approach in communication channel analysis. This study contributes to the advancement of research in radio propagation at the 3.5 GHz frequency by combining simulation, measurements, and artificial intelligence techniques for a robust and accurate analysis.ptAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/Pequena e larga escalaRMSEModelos de perda de percursoRede neuralSmall-scale and large-scalePath loss modelsNeural networkDissertaçãoCNPQ::ENGENHARIAS::ENGENHARIA ELETRICAINTELIGÊNCIA COMPUTACIONALCOMPUTAÇÃO APLICADA