Resumen: A Video Surveillance System (VSS) is of primary importance in public transport hubs such as airports, train or bus stations but also inside the vehicle itself. In this paper, we present a heuristic architecture model for on-board video surveillance system based on Internet of Video Things (IoVT) devices which addresses the need for delivering smart video surveillance in public transport vehicles (e.g., buses) minimizing the impact on network performance. A proof-of-concept was implemented using a public Cloud Service Provider (CSP) and two Raspberry Pi as edge computing nodes. On the edge nodes, a Machine Learning (ML) application was deployed along with a network-efficient video streaming system. On the other hand, laboratory tests are included to understand the network traffic dynamic, furthermore results are enhanced with a set of simulations in order to analyze the performance of a video streaming application and 6 different congestion control algorithms in terms of packet loss and delay. Idioma: Inglés DOI: 10.1109/TLA.2021.9477277 Año: 2021 Publicado en: IEEE LATIN AMERICA TRANSACTIONS 19, 10 (2021), 1763-1771 ISSN: 1548-0992 Factor impacto JCR: 0.967 (2021) Categ. JCR: ENGINEERING, ELECTRICAL & ELECTRONIC rank: 244 / 274 = 0.891 (2021) - Q4 - T3 Categ. JCR: COMPUTER SCIENCE, INFORMATION SYSTEMS rank: 152 / 163 = 0.933 (2021) - Q4 - T3 Factor impacto CITESCORE: 2.1 - Computer Science (Q3) - Engineering (Q3)