ReG3M: reencaminhamento geográfico de mensagens multimídia homográficas em congestionamentos veiculares

• Autor(a) principal: Cunha Júnior, Lair Sérvulo da
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/43293
• topic: Broadcast storm; Redes veiculares; Disseminação de mensagens; Monitoramento por vídeo; Qualidade de imagem; Homografia; Vehicular networks; Message dissemination; Video monitoring; Image quality; Homography; Ciência da Computação

The use of technology has become increasingly popular, with this popularization, several sectors started to use as technologies, such as vehicular networks. Vehicle networks or VANETs, are networks formed by the communication between own vehicles (V2V), between vehicles and infrastructure (V2I) and hybrid (V2X). Within VANETs there are several types of applications, one of which seeks to improve traffic efficiency. One way to improve traffic efficiency is to use videos to monitor roads. In situations of congestion, videos can be useful to inform the state in which if the incident that caused the congestion occurs. One way to send the video to as many people as possible is to send by broadcast, however this type of sending can cause problems. To avoid problems caused by sending in broadcast, this work proposes a method of Geographical Forwarding of MultiMedia Messages (ReG3M). ReG3M uses the vehicle farthest from the message source to be the forwarder, it also controls repeated messages on the network. ReG3M was compared to transmission in broadcast and the AID and DBRS algorithms. This work also offers a Framework for using video homography in vehicular networks (VANET) in congestion scenarios. The SIFT, SURF, ORB and BRISK algorithms were obtained and compared when using the RANSAC algorithm to create the homographic matrix, considering the video quality algorithms BRISQUE and NIQE. Simulation results that BRISK obtained better image quality in relation to the other descriptors being also the second fastest in processing time. ReG3M was more efficient in almost all tests, in sending messages than the other algorithms, reaching more vehicles and a greater distance in less time. Being better mainly in messages with larger series sizes or in case of videos created by the homography process