Loading...
Publication
Otimização da entrega over-the-air em veículos conectados
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.3 MB | Adobe PDF |
Advisor(s)
Abstract(s)
A evolução dos veículos conectados trouxe novos desafios às comunicações over-the-air (OTA),
que exigem soluções fiáveis, escaláveis e eficientes. Esta dissertação, elaborada no âmbito da
Unidade Curricular de Dissertação do Mestrado em Engenharia Informática do Instituto
Superior de Engenharia do Porto teve como objetivo estudar e otimizar a entrega de conteúdo
OTA através da análise de protocolos de comunicação, algoritmos diferenciais e estratégias de
compressão e de cache, focando-se na minimização da latência e do tamanho dos ficheiros
enviados (payload).
Foi desenvolvida uma arquitetura experimental composta por um backend em Java 21 com
Spring Boot e um cliente Raspberry Pi 5, permitindo simular cenários realistas de atualização de
blobs de mapas. Foram implementados e avaliados algoritmos diferenciais (bsdiff e xdelta3),
combinados com compressão (gzip) e com o conceito de hot zones em cache. As métricas
analisadas incluíram tempo de resposta end-to-end e tamanho de payload transferido.
Os resultados demonstraram que o uso de pacotes de atualizações (patches) diferencias
reduziram substancialmente a latência (até 75%) e o volume de dados transmitidos (cerca de
90%), em comparação com o envio de blobs completos. Verificou-se que xdelta3 é mais
eficiente em tempo, enquanto bsdiff gera patches ligeiramente menores. A compressão com
gzip apenas se mostrou benéfica em artefactos de maior dimensão, sendo contraproducente
em patches pequenos. A estratégia de hot zones mostrou-se essencial para garantir
escalabilidade, reduzindo em cerca de 50% o tempo médio de resposta em cenários de elevada
reutilização de conteúdos.
Este estudo conclui que a combinação de algoritmos diferenciais, compressão seletiva e cache
baseada em hot zones constitui uma abordagem eficaz para otimizar comunicações OTA. São
ainda propostas perspetivas de evolução que incluem a integração de outros protocolos (MQTT,
QUIC), simulação de redes adversas e políticas adaptativas de compressão e envio.
The evolution of connected vehicles has introduced new challenges to over-the-air (OTA) communications, which require reliable, scalable, and energy-efficient solutions. This dissertation, developed within the Master’s in Computer Engineering at Instituto Superior de Engenharia do Porto, aimed to study and optimize OTA content delivery by analyzing communication protocols, differential algorithms, and strategies for compression and caching. An experimental architecture was developed, consisting of a backend built in Java 21 with Spring Boot and a client running on a Raspberry Pi 5, enabling the simulation of realistic map blob update scenarios. Differential algorithms (bsdiff and xdelta3) were implemented and evaluated, in combination with compression (gzip) and a caching strategy based on hot zones. The main metrics considered were end-to-end response time and payload size. The results showed that using differential patches significantly reduces latency (up to 75%) and data volume (around 90%) compared to full blob delivery. It was observed that xdelta3 achieves better performance in terms of response time, while bsdiff generates slightly smaller patches. gzip compression proved beneficial only for larger artefacts, being counterproductive for small patches. The hot zone caching strategy was essential to scalability, reducing average response times by approximately 50% in scenarios with high content reuse. This study concludes that the combination of differential algorithms, selective compression, and hot zone caching provides an effective approach to optimizing OTA communications. Future work should explore the integration of additional protocols (MQTT, QUIC), the simulation of adverse network conditions, and adaptive policies for compression and transmission.
The evolution of connected vehicles has introduced new challenges to over-the-air (OTA) communications, which require reliable, scalable, and energy-efficient solutions. This dissertation, developed within the Master’s in Computer Engineering at Instituto Superior de Engenharia do Porto, aimed to study and optimize OTA content delivery by analyzing communication protocols, differential algorithms, and strategies for compression and caching. An experimental architecture was developed, consisting of a backend built in Java 21 with Spring Boot and a client running on a Raspberry Pi 5, enabling the simulation of realistic map blob update scenarios. Differential algorithms (bsdiff and xdelta3) were implemented and evaluated, in combination with compression (gzip) and a caching strategy based on hot zones. The main metrics considered were end-to-end response time and payload size. The results showed that using differential patches significantly reduces latency (up to 75%) and data volume (around 90%) compared to full blob delivery. It was observed that xdelta3 achieves better performance in terms of response time, while bsdiff generates slightly smaller patches. gzip compression proved beneficial only for larger artefacts, being counterproductive for small patches. The hot zone caching strategy was essential to scalability, reducing average response times by approximately 50% in scenarios with high content reuse. This study concludes that the combination of differential algorithms, selective compression, and hot zone caching provides an effective approach to optimizing OTA communications. Future work should explore the integration of additional protocols (MQTT, QUIC), the simulation of adverse network conditions, and adaptive policies for compression and transmission.
Description
Keywords
Over-the-air veículos conectados V2X algoritmos diferenciais compressão caching compression differential algorithms connected vehicles